Glen Gowers, the Co-Founder and CEO of Basecamp Research, emphasizes the potential of artificial intelligence to revolutionize the field of biological research. He argues for the development of AI systems that possess an innate understanding of biological concepts, which could facilitate groundbreaking discoveries in the life sciences. With the right technological advancements, AI could enhance our ability to unravel complex biological processes and lead to significant breakthroughs across various research sectors.

Basecamp Research: We require AI that naturally comprehends biology

A sunlit, modern biology lab where a sleek AI core bearing the Basecamp logo sits next to a curved translucent touchscreen showing a living, rotating 3D DNA helix; two adult researchers in white lab coats lean in, taking notes and exchanging hopeful smiles; warm daylight streams through floor-to-ceiling windows onto microscopes, petri dishes, and open biology journals, evoking intuition and collaboration.

Basecamp Research: We need AI that intuitively understands biology - CNBC

- Glen Gowers emphasizes the importance of AI in advancing biological research discoveries.  
- AI should intuitively comprehend biological concepts for effective research applications.  
- Basecamp Research aims to leverage AI technology for enhanced scientific insights.

Share Share Article via Facebook Share Article via Twitter Share Article via LinkedIn Share Article via Email

Basecamp Research: We need AI that intuitively understands biology

Glen Gowers, Co-Founder and CEO of Basecamp Research, discusses how AI can be harnessed to help make new discoveries in biological research.

Basecamp Research: We need AI that intuitively understands biology

New features are being introduced for AI Max, empowering advertisers with enhanced control and insightful reporting to better grasp performance metrics.

To start, we are expanding the closed beta for AI Brief to include more languages, such as Dutch, French, German, Italian, Japanese, Portuguese, and Spanish. AI Brief allows advertisers to direct AI Max using their own language, providing comprehensive context about their business, target audience, and essential messaging.

Additionally, a new reporting feature is making its way into Google Ads, designed to offer a clearer perspective on AI-driven performance. This feature will present a cohesive overview of the Search ads journey, enabling you to easily identify which search terms activated your ads, the specific creative assets that users encountered, and the exact landing pages on your website. This detailed visualization will enhance your understanding of customer interactions with your business while highlighting the strategic advantages that AI Max contributes to your advertising efforts.

Further details and availability updates will be made available later this year.

We're expanding AI Brief to additional languages and introducing a new AI Max reporting feature.

A bright, sunlit glass-walled innovation lab with natural light flooding the space. In the center sits a sleek circular digital table displaying a glowing world map overlaid with multi-language snippets—“Hola,” “你好,” “Bonjour,” “مرحبا”—and two side panels labeled “AI Brief” and “AI Max Reporting.” A diverse team of professionals in business attire stands around the table, pointing and discussing insights. Crisp “OpenAI” text is etched onto the table’s base with no logo. The scene conveys global expansion and powerful new reporting features.

We’re bringing AI Brief to more languages and adding a new AI Max reporting feature. - blog.google

- New AI Max features enhance advertiser control and performance reporting capabilities.  
- AI Brief expands to multiple languages, allowing personalized guidance for advertisers.  
- Google Ads introduces unified reporting for better insights on AI-driven ad performance.  

We’re rolling out new AI Max features that provide advertisers with greater control and new reporting to understand performance.

First, we’re bringing the closed beta of AI Brief to more languages: Dutch, French, German, Italian, Japanese, Portuguese, and Spanish. AI Brief lets you guide AI Max in your own words with detailed context on your business, audience, and key messaging.

Second, we’re introducing a new reporting feature in Google Ads to give you a clearer view of AI-driven performance. It will show a single, unified view of the Search ads journey — making it easy to see what search terms triggered your ads, which creative assets the user saw, and exactly where they landed on your website. This precise interface will help you better understand how customers interact with your business and validate the strategic value AI Max brings to your campaigns.

We’ll share additional details and availability later this year.

We’re bringing AI Brief to more languages and adding a new AI Max reporting feature.

According to the OECD's latest economic forecast, the global economy is expected to grow by 2.9% in 2026, showing a slight improvement from the previously estimated 2.8% in June. For 2027, the growth projection has been revised down to 3.0% from an earlier forecast of 3.1%. The organization highlights that while investments in artificial intelligence (AI) have been somewhat supportive of the global economy this year, ongoing energy crises and geopolitical tensions are dampening the outlook for the coming years.

Following a robust 3.4% growth in the previous year, the deceleration is attributed not only to the persistent energy challenges but also to the impact of commodity price fluctuations tied to conflicts in the Middle East. The OECD emphasizes that the surge in AI spending, particularly in infrastructure such as data centers and semiconductors, has been a vital factor in maintaining some economic resilience, particularly in the United States, as well as bolstering tech exports from countries like Japan and South Korea.

However, the OECD cautions that various risk factors could disrupt this optimistic scenario. Potential volatility in energy markets, the effects of a strong El Niño, rising government bond yields, and underwhelming returns from AI investments could collectively lower global growth by as much as 0.7 percentage points next year and inflate global prices by 1.1 percentage points.

Turning to inflation forecasts, the OECD projects that inflation within G20 nations will rise to 4.1% in 2026, adjusted upward from 4.0% in June, and further to 3.6% in 2027, which is notably higher than the previous estimate of 3.1%. This inflationary pressure may compel central banks to consider interest rate adjustments if broader price hikes or slowed growth trends emerge.

Focusing on individual regions, the United States is poised for growth of 2.2% this year and 2.1% in 2027, both figures being upgrades from previous projections. Heavy investment in AI is expected to offset lower consumer spending levels, while U.S. inflation is anticipated at 3.6% in 2026, decreasing to 2.6% in 2027 amidst rising tariffs and energy costs impacting consumer spending power.

China's growth rate is predicted to decrease to 4.5% this year and 4.2% in 2027, consistent with earlier forecasts. This slowdown is largely a result of government measures aimed at curbing excessive industrial capacity, despite a gradual recovery in consumption.

In the Euro zone, growth is expected to remain at 1.0% for both 2026 and 2027, hindered by high energy prices and interest rates, although new defense spending initiatives may provide some support. Inflation in the Euro zone is projected at 3.0% for this year and 2.9% for 2027, partly driven by rising natural gas prices amid low storage levels ahead of winter.

For Japan, economic growth is forecast at 0.8% in 2026 and 0.7% in 2027, with heightened policy rates and rising energy import costs impacting growth despite notable business investments. Japan's inflation is anticipated to rise to 2.6% in 2027, driven by a tight labor market and increasing wages.

Canada's growth estimates have been adjusted downward, with a 2026 forecast of 0.9%, down from 1.2%, and a 2027 outlook of 1.3%, reduced from 1.7%, mainly due to new U.S. tariffs affecting Canadian exports. 

This summary provides a comprehensive view of projected economic trends as outlined in the OECD's latest report, reflecting a mix of resilience fueled by AI investment and challenges posed by energy prices and geopolitical conflicts.

OECD anticipates that the AI surge will alleviate the impact of the energy crisis in the Middle East for the time being.

Warm sunrise over a desert oil field: a single pumpjack with a fluttering Saudi Arabian flag. Beside it, a shipping-container style data center with its doors open to rows of servers softly humming, green status lights dancing. In the foreground, a wooden crate stacked high with papers stamped “OECD” showing upward-trending graphs and oil barrel icons. Soft natural light bathes the scene, evoking hope that an AI boom can help offset the Middle East energy shock.

OECD expects AI boom to help offset Middle East energy shock for now - Reuters

- Global growth is projected to reach 2.9% in 2026 and 3.0% in 2027.  
- G20 inflation is forecasted at 4.1% in 2026 and 3.6% in 2027.  
- Risks like El Niño and AI earnings could reduce growth by 0.7 percentage points.  

Summary Global growth seen at 2.9% in 2026 (vs 2.8% in June), 3.0% in 2027 (vs 3.1%)

G20 inflation forecast at 4.1% in 2026, revised up sharply to 3.6% in 2027

Risks from El Niño, bonds to AI earnings could cut global growth by 0.7 points

PARIS, Sept 23 (Reuters) - AI-led investment is helping the global economy hold up marginally better than expected this year, but the ‌energy shock is becoming more entrenched, weighing on the outlook for 2027, the OECD said on Wednesday.

After 3.4% growth last year, the global economy is set to slow to 2.9% growth in 2026, slightly better than the 2.8% forecast in June, the Organisation for Economic Co-operation and Development said in its interim ​economic outlook.

Sign up here.

Heading into 2027, the commodity price shock caused by the Middle East conflict is expected to weigh on momentum ​and the OECD forecasts global growth picking up to only 3.0%, from 3.1% in June.

The OECD said ⁠strong spending on AI infrastructure, from data centres to semiconductors, has been a key pillar of resilience this year, boosting growth ​in the United States and lifting technology exports from Japan and Korea.

However, it warned the global outlook was particularly clouded by the potential ​for energy market jitters, extreme weather related to a strong El Niño, surging government bond yields and disappointing AI investment returns.

If those risks materialised, the OECD estimated they could together reduce global growth by 0.7 percentage points next year and raise global inflation by 1.1 percentage points.

In the OECD's baseline outlook, ​inflation in G20 economies was seen at 4.1% in 2026, up from 4.0% forecast in June. The OECD also raised its 2027 ​forecast to 3.6%, from 3.1% in June, which it said could force central banks to adjust interest rates if price pressures broaden out or growth ‌falters.

DIVERGENT OUTLOOKS

In ⁠the US, the world's largest economy, growth is seen at 2.2% this year and 2.1% in 2027, both upgrades from June, as heavy AI-related investment offsets weaker consumer spending. US inflation is projected to hit 3.6% in 2026, easing to 2.6% in 2027 with tariffs and higher energy prices weighing on household purchasing power and business costs.

China's growth is expected to slow to 4.5% this year and 4.2% ​in 2027, unchanged from June, ​as Beijing's curbs on excess ⁠industrial capacity weigh on investment even as consumption faces a gradual pick up in inflation.

Euro zone growth is seen holding at 1.0% in both 2026 and 2027, with higher energy prices and interest rates ​weighing on activity before new defence spending initiatives provide support.

Euro zone inflation is forecast at 3.0% ​this year and ⁠2.9% in 2027, driven partly by a jump in natural gas prices as European storage levels sit at 15-year lows heading into the winter heating period.

Japan's economy is expected to grow 0.8% in 2026 and 0.7% in 2027, with rising policy rates and costlier energy imports offsetting ⁠strong business ​investment. Unlike other major economies, Japan's inflation is seen accelerating to 2.6% in ​2027 from 1.8% this year, reflecting a tight labour market and strong wage growth.

Canada's 2026 growth forecast was cut to 0.9% from 1.2% in June while its 2027 ​outlook was lowered to 1.3% from 1.7% due to the new US tariffs on Canadian exports.

Reporting by Leigh Thomas; Editing by Alexander Smith

Our Standards: The Thomson Reuters Trust Principles. , opens new tab

OECD expects AI boom to help offset Middle East energy shock for now

Upon clicking the “Proceed” button below, you will be taken to a separate website that is operated by Hong Kong Tiimoot Information Technology Co., Limited (HKT), situated in Hong Kong. Please note that this site functions independently of Barron's, which means that we do not have oversight or control over its content. It’s important to review HKT’s Privacy Statement, as their practices regarding privacy and data may differ from those of Barron's. We cannot be held accountable for the privacy practices or data handling methods of HKT.

Muse AI Elevates Expectations for Meta Connect Keynote

A grand conference hall bathed in warm, natural light. Center stage features a massive screen showcasing the Muse AI logo in bold relief, its emblem rising like towering stakes. To the right, a sleek podium with the Meta logo stands ready. In the foreground, silhouetted delegates gaze upward, anticipation crackling in the golden glow.

Muse AI Has Raised Stakes for Meta Connect Keynote - Barron's

- Clicking “Proceed” redirects you to a third-party site operated by HKT.  
- HKT operates independently from Barron's and is based in Hong Kong.  
- Barron's is not liable for HKT's privacy or data practices.  

You are now leaving Barron's website

By clicking on the “Proceed” button below, you will be redirected to a third-party website owned and operated by Hong Kong Tiimoot Information Technology Co., Limited. (“HKT”), which is located in Hong Kong. That website operates independently from Barron's and Barron's does not control the website. The privacy practices of HKT are subject to its Privacy Statement, so please read it closely. We are not responsible for HKT's privacy or other data-related practices.

Muse AI Has Raised Stakes for Meta Connect Keynote

Your privacy preferences

When you select 'Accept all,' we and our partners, including 254 members of the IAB Transparency & Consent Framework, will collect and utilize technical identifiers along with your browsing and search data. This information is employed for analytics, personalized advertising, content measurement, audience research, and the development of services. By clicking 'Accept all,' you are permitting the use of cookies, precise geolocation data, and other personal information for these various purposes. 

If you prefer to opt-out of the use of cookies and personal data for these additional aims, please click 'Reject all.'

For those who wish to tailor their privacy options, you can click 'Manage privacy settings.'

You have the flexibility to revoke your consent or alter your preferences whenever you like by visiting the 'Privacy and Cookie settings' or 'Privacy dashboard' links available on our websites and applications. To learn more about the ways we utilize your personal data, please refer to our Privacy Policy and Cookie Policy.

Your options for privacy

A sunlit home office bathed in warm morning light, where a focused adult sits at a wooden desk opening “Privacy Settings” on their laptop screen—rows of colorful on/off toggles and glowing lock icons align like stepping stones toward a set of real padlocks arranged on the desktop, symbolizing empowered choices and control.

AI image of Australia PM's arrival in San Jose airport dupes critics - yahoo.com

- Users can accept or reject cookies for personalized advertising and content.  
- Customization of privacy choices is available through 'Manage privacy settings' option.  
- Consent can be withdrawn anytime via 'Privacy and Cookie settings' links.

Your privacy choices

Accept all', we and technical identifiers and browsing and search data, for analytics, personalised advertising and content, advertising and content measurement, and audience research and services development. If you click '', we and our partners , including 254 who are part of the IAB Transparency & Consent Framework, will also store and / or access information on a device (in other words, use cookies) and use precise geolocation data and other personal data such asand browsing and search data, for analytics, personalised advertising and content, advertising and content measurement, and audience research and services development.

If you do not want us and our partners to use cookies and personal data for these additional purposes, click 'Reject all'.

If you would like to customise your choices, click 'Manage privacy settings'.

You can withdraw your consent or change your choices at any time by clicking on the 'Privacy and Cookie settings' or 'Privacy dashboard' links on our sites and apps. Find out more about how we use your personal data in our Privacy Policy and Cookie Policy.

Your privacy choices

When Xi Jinping, leader of China, arrives in Maryland on Wednesday, he will be greeted by a warm reception—possibly even a hug, according to the hopes of Donald Trump—marking a notable shift in US-China relations under the current leadership compared to recent administrations.

This meeting will be the second face-to-face encounter this year between the leaders of the world's two largest economies, coming after their initial discussions in May in Beijing. During that summit, both leaders expressed a desire to foster "strategic stability" between their countries; however, tangible results were lacking, with ongoing disputes over trade, export limitations, and geopolitical matters persisting.

Despite the friendly atmosphere from the earlier meeting, expectations are muted for meaningful advancements during Xi's first official visit to the US since 2015. Bonnie Glaser, managing director at the German Marshall Fund, stated, “Expectations are very low,” noting that the term “deliverables” is rarely being mentioned.

One potential area for agreement between Trump and Xi could be artificial intelligence safety. A number of prominent tech leaders, including Jeff Bezos, Sundar Pichai, Sam Altman, Tim Cook, Elon Musk, Mark Zuckerberg, and Jensen Huang, are expected to be present at a state dinner at the White House on Thursday night.

Trump has recently voiced concerns about the rapid advancement of AI and emphasized the importance of the US maintaining its lead over China in this critical technology sector. “We’re leading China in AI,” he remarked. “Frankly, I want to keep it that way, because whoever wins AI, wins.”

This past Sunday, Scott Bessent, the US Treasury Secretary, met with Chinese Vice Premier He Lifeng and proposed establishing an AI safety framework to strengthen collaboration between Washington and Beijing. However, this proposal appears limited in scope, lacking a major safety agreement.

In a recent article, China's state-run People’s Daily advocated for a cooperative approach to AI, suggesting it as a promising avenue for US-China collaboration. António Guterres, the UN Secretary-General, echoed these sentiments, urging both nations to create channels for dialogue akin to those seen during the Cold War era.

Nevertheless, China contests US claims regarding the existential threat posed by its AI advancements. The Chinese foreign ministry pushed back against an essay that advocated for US restrictions on Chinese AI development, asserting that “fearmongering, confrontation, and vicious competition” would only hinder global AI governance efforts.

Beijing has taken measures to regulate AI development, including requiring the labeling of AI-generated content and placing restrictions on children’s use of AI companions, in alignment with the Chinese Communist Party's goals of information control.

The trade issue remains ever-present, with a temporary truce in the trade war initiated last year during talks in South Korea set to expire in November. Beijing desires a long-term suspension of tariffs that threaten its export-reliant economy, but the US is reportedly hesitant to agree to terms that might undermine its negotiating power.

Glaser pointed out, “The US side views this as leverage over China and ensuring that they can get deliveries of the rare earths and magnets that American companies need.” Despite the ongoing trade tensions, China’s exports have flourished, heading toward a trade surplus of $1 trillion for the second consecutive year.

Another significant topic that has evolved since their last meeting is Taiwan. Following the May summit, Xi gained a key concession when Trump agreed to temporarily halt a significant package of arms sales to Taiwan—an issue traditionally excluded from US-China discussions. However, Trump has since labeled future arms sales as a “very good negotiating chip,” indicating a potential delay in such sales until after upcoming APEC and G20 summits.

In light of the ongoing geopolitical landscape, experts are cautioning against expectations for significant progress in this summit. Scott Kennedy from the Center for Strategic and International Studies remarked that the US-China relationship is not evolving into a cooperative “G2” dynamic, but rather a more complex engagement strategy.

China remains a leading player in oil trade with Iran, and while it has put pressure on Tehran to negotiate in light of US strikes, its overall influence in mediating the Middle East crisis has been limited. Similarly, China's unwavering support for Russia throughout the ongoing Ukraine conflict emphasizes its position as a stabilizing force in contrast to the US’s ongoing challenges in the region.

With the divergent approaches of both nations, the upcoming summit is expected to yield minimal concrete outcomes. Amanda Hsiao suggested that the talks will maintain a façade of cordiality but lack substantive ambition.

Democratic voices in Washington have expressed concerns over the domestic implications of Trump’s potentially conciliatory stance toward Xi, with Senator Tammy Duckworth highlighting the need for the president to address human rights issues in China seriously. This sentiment illustrates anxieties that Trump's diplomatic overtures may overshadow pressing concerns regarding Chinese coercive practices.

AI casts a shadow over Trump-Xi meeting as distrust between the US and China deepens.

Cinematic digital art, a dimly lit conference room with two empty ornate chairs facing each other across a polished mahogany table. On the left wall hangs a U.S. flag, on the right a Chinese flag. Floating above the table is a giant stylized neural-network brain glowing in ominous red and orange, casting deep shadows across scattered papers and microphones. Dramatic low-key natural light highlights the tension and distrust in the air.

AI looms large over Trump-Xi meeting amid deep distrust between US and China - theguardian.com

- Xi Jinping's visit marks his first state trip to the US since 2015.  
- Expectations for significant outcomes from the Trump-Xi meeting are notably low.  
- Trade negotiations are complicated by differing views on tariffs and AI safety.  

When Xi Jinping, China’s leader, touches down in Maryland on Wednesday, he will step out into the arms – perhaps a hug, if Donald Trump’s wishes come true – of a US president who has shown China more bonhomie than any of his recent predecessors.

The meeting between Xi and Trump this week will be the second time that the leaders of the world’s two biggest economies have talked face to face this year. At the previous summit, held in Beijing in May, there was much talk of building a “strategic stability” between the two powers. But there was little by way of concrete outcomes: the countries remain at loggerheads over trade, export controls and geopolitics.

Despite the goodwill built in May, few anticipate any major breakthroughs this week when Xi makes his first state visit to the US since 2015. “Expectations are very low,” said Bonnie Glaser, a managing director at the US thinkthank the German Marshall Fund. “Nobody is using the term ‘deliverables’.”

One area in which Trump and Xi may reach some consensus is on artificial intelligence safety. Tech executives including Jeff Bezos of Amazon, Sundar Pichai of Alphabet, Sam Altman of ⁠OpenAI, Tim Cook of Apple, Elon Musk ​of ​Tesla, Mark Zuckerberg of Meta and Jensen Huang ​of Nvidia will reportedly ​attend a ⁠state dinner at the White House on Thursday evening.

‘Whoever wins AI, wins’

In the past few weeks, the US has been beset by worries that AI is progressing too fast, but that the US must strive to defend its lead over China in the frontier technology – and Trump has pushed back hard against calls to slow its development. “We’re leading China in AI,” Trump said recently. “And frankly, I want to ⁠keep it that way, because whoever wins AI, wins.”

On Sunday, Scott Bessent, the US treasury secretary, concluded talks with the Chinese vice-premier He Lifeng by proposing an AI safety mechanism to be established between Washington and Beijing. The mechanism appears to stop short of a major safety agreement.

View image in fullscreen Scott Bessent, the US treasury secretary, and He Lifeng, Chinese vice-premier, during trade talks in Paris, France, on 16 March. Photograph: US treasury/Reuters

Tuesday’s front page of the People’s Daily, the official newspaper of the Chinese communist party (CCP), ran an article saying that the US and China should “make AI a new frontier for China-US cooperation” and “make good use of the intergovernmental dialogue mechanism on AI”.

António Guterres, the UN secretary general, called on the two countries to establish a dialogue on AI similar to the communication between the US and the Soviet Union during the cold war. “Governments with the greatest AI capabilities have the greatest responsibilities to humanity, to establish channels for dialogue, transparency, trust and cooperation,” Guterres said.

Still, China does not agree with US assessments about the grave threat posed by its AI developments. China’s foreign ministry responded to an essay by Dario Amodei, the Anthropic chief executive, that called for Washington to impede Beijing’s progress, by saying “fearmongering, confrontation and vicious competition will only disrupt the process of global AI governance”.

China has already taken steps to mandate the labelling of AI-generated content, to restrict the ability of children to use AI companions and, in keeping with the CCP’s desire to keep tight control on all information, to restrict what can be said by chatbots.

China may be more concerned with what it sees as US attempts to limit is growth. Washington has for years restricted China’s ability to buy the most advanced semiconductors needed for AI development, and this year the US banned the sale of Chinese humanoid robots.

Amanda Hsiao, China director at the Eurasia Group, a political risk consultancy, said: “China doesn’t necessarily see the threat in the same way that the US does. There is a lot of distrust that’s in the way of cooperation. The Chinese are inclined to think of every US call for safety standards or calls from companies for pacing the frontier as means of suppressing Chinese development.”

Trade tensions persist

Next on the agenda will be trade. Washington and Beijing struck a temporary truce in the trade war last year at a summit in South Korea. That truce – which included a pause on tariffs, as well as an agreement for Beijing to loosen restrictions on the export of rare earths – will expire in November.

Beijing is hoping for a longer-term commitment to banish tariffs, which pose a threat to China’s export-driven economy. But reports suggest that the US is unwilling to agree to anything that would weaken its future bargaining power.

Glaser, of the German Marshall Fund, said: “The US side views this as leverage over China and ensuring that they can get deliveries of the rare earths and magnets that American companies need.”

China may not feel the need to negotiate, either. Despite the trade war, China’s exports have surged, and it is on track to reach a $1tn trade surplus for the second year running.

‘This is not a G2’

One aspect of the relationship that has shifted since May is the issue of Taiwan. After the Beijing summit, Xi secured a major concession from Trump, when he agreed to pause a $14bn package of US arms sales to Taiwan. US presidents have traditionally excluded the issue of arms sales to Taiwan, a self-governing island that China claims as part of its territory, from the US-China relationship.

But Trump has described future weapons sales as a “very good negotiating chip”. The US is now reportedly considering delaying further sales until after the Apec and G20 summits later this year, where Trump and Xi are expected to meet again.

Tuesday’s People’s Daily article said that a stable China-US relationship would “bring more peace, prosperity, and progress to the world”. But for all the talk in Beijing of “strategic stability”, Scott Kennedy, a senior adviser at the Center for Strategic and International Studies, notes that “this is not a G2” alliance of the world’s two dominant powers.

View image in fullscreen Trump and Xi Jinping tour the Forbidden City in Beijing, China, on 8 November 2017. Photograph: Andrew Harnik/AP

“They are finding a way to manage the bilateral relationship, but they are not whatsoever working together as co-stewards of the international order,” Kennedy wrote in a recent commentary.

China is Iran’s biggest oil buyer and trading partner. After Trump launched strikes against Iran in February, China reportedly put pressure on Tehran behind the scenes to come to the negotiating table. It also slashed oil imports by nearly one-quarter in the first six months of the war, limiting global inflation.

Da Wei, a professor of international relations at Tsinghua University, wrote in a recent essay that this reflected “a more confident China” that “did not exploit the difficulties that the United States faced in the war with Iran”. But Beijing has otherwise done little to resolve the crisis in the Middle East, which it sees as being a mess of Trump’s making.

Last month, Bessent announced an “economic D-day” against Iran and its enablers. The US treasury secretary subsequently claimed to have had “very good private discussions” with Chinese officials over ending the country’s economic ties with Iran.

When it comes to Ukraine, Beijing has shown unwavering support for Moscow since Russia’s president, Vladimir Putin, launched his full-scale invasion in 2022.

Ali Wyne, senior research and advocacy adviser on the US and China at the International Crisis Group, said: “The contrast between the world’s incumbent power and its principal challenger will be even sharper in Washington than it was in Beijing in May: while the former is mired once more in the Middle East and straining longstanding alliances, the latter is casting itself as a stabilizing counterweight and the leading champion of a more multipolar order.”

High vibes, low ambition

The lack of momentum on thorny geopolitical issues, combined with China’s emphasis on maintaining “strategic stability” above all else, means that few expect there to be major breakthroughs at this week’s summit.

Hsiao added: “It is going to be yet another summit that is high on positive vibes and relatively low ambition in terms of substance.”

Democrats in Washington have urged Trump to raise concerns over China’s treatment of political prisoners, but are already sounding the alarm that he could make too many concessions to Xi.

Senator Tammy Duckworth told an audience at the Center for American Progress thinktank on Tuesday: “He’s not very good at making deals with the PRC [People’s Republic of China] that they actually follow through on. If you look at the result of the last summit they had, they promised the purchase of 200 Boeing aircraft and $17bn in agricultural products. Neither one of those has come to pass.”

“At the same time, he violated our longstanding position on Taiwan by engaging in discussions on arms sales to Taiwan with President Xi,” Duckworth added. “I worry that it’s going to be a repeat of that and President Trump basically spends too much time sucking up to President Xi. He’s very distracted. When he focuses on the Indo-Pacific, it’s about him cosying up to an authoritarian regime at the expense of our friends and allies in the region.”

Additional research by Yu-chen Li

AI looms large over Trump-Xi meeting amid deep distrust between US and China

Consumers are beginning to embrace the potential of personal artificial intelligence, with many placing their confidence in Muse, Meta's latest AI assistant. The app has quickly climbed to the top of the Apple App Store rankings, surpassing the widely-used ChatGPT, reflecting a growing trend of individuals relying on AI to enhance their shopping experiences and save them both time and money, despite wider concerns about AI's implications.

Neil Saunders, the managing director and retail analyst at GlobalData Retail, highlighted the excitement surrounding Muse, which he believes signifies a shift towards more proactive shopping experiences. He anticipates that as consumers seek to simplify tedious tasks—like comparing prices across various platforms and locating products—they will increasingly turn to AI shopping assistants like Muse.

While Saunders hesitates to declare this as a definitive turning point, he acknowledges that it represents a significant phase in the evolving relationship between AI applications and consumers, as he shared with CBS News.

Research from IBM indicates that 41% of shoppers currently utilize AI assistants for product research, with 31% leveraging them to hunt for discounts. Since Muse's launch just two weeks ago, it has reportedly garnered over 2.5 million downloads, according to data shared by Sensor Tower. In comparison, OpenAI's ChatGPT app achieved 3.1 million downloads during the same timeframe, while Anthropopi’s Claude managed about 200,000 downloads.

Further emphasizing Muse's relevance, Amazon recently informed CBS News that it has prohibited the app from operating on its site due to concerns regarding security and user experience.

The landscape of online search is changing. For a long time, consumers used AI mainly as an advanced search engine. However, Meta is positioning Muse as a practical tool designed to assist users with various everyday tasks, aiming to enhance convenience in their lives.

John Mercer, head of global research at Coresight, noted that if Muse continues to gain traction, it may surpass Claude in popularity and become one of the leading AI tools among users. Unlike conventional chatbots, Muse's capabilities extend to executing tasks on users' behalf, such as booking travel, organizing email, and monitoring price changes. This not only means consumers can save time but also aids them in unearthing the best deals.

Mercer described this advancement as a "natural evolution from search" functionality. A recent survey by Coresight involving over 1,000 consumers found that 30% of participants expressed interest in using an AI tool to compare prices while holiday shopping, with the same percentage willing to use such tools to discover deals or promotional codes.

Saunders explained that agentic commerce aims to curate and display personalized selections for users, effectively minimizing the effort required to browse multiple websites for comparisons. 

While Muse is already causing a stir, Saunders cautions that a wider acceptance of AI shopping agents is likely to take some time. Reports of AI-related mishaps have raised concerns, making consumers hesitant to trust these agents with purchasing decisions.

"There's still a level of apprehension surrounding the idea of allowing AI agents to manage purchases on their behalf," he remarked.

Meta's Muse personal assistant gains popularity as AI-driven commerce begins to flourish.

Bright midday sun streams through a modern living room window onto a sleek tablet resting on a coffee table. On its screen, a friendly AI shopping assistant interface offers product suggestions. Nearby, neatly stacked delivery boxes hint at ongoing online orders. An adult’s hand reaches toward the tablet, as soft natural light highlights a large Meta logo embossed on the device. The scene conveys effortless AI-powered commerce taking off.

Meta's Muse personal agent is a hit, as AI commerce starts to take off - CBS News

- Meta's Muse AI agent leads the App Store, surpassing ChatGPT in popularity.  
- 41% of consumers use AI assistants for product research and deal searching.  
- Amazon has blocked Muse from its platform due to security and user experience concerns.  

Shoppers have seen the future, and many of them have decided it looks like Muse.

Meta's new personal artificial intelligence agent now sits atop Apple's App Store, ahead of No. 2 ChatGPT, as more consumers put their faith in AI to save them time and money — existential AI fears be damned.

Neil Saunders, managing director and retail analyst at GlobalData Retail, said the buzz around Muse underscores the "acceleration of agentic shopping." He expects more people to gravitate to AI shopping agents like Muse as they look to hand off cumbersome tasks, such as comparing deals across sites and tracking down products.

"I wouldn't say it's a tipping point, but it's certainly an important part of the journey for AI and consumers," Saunders told CBS News.

According to IBM, 41% of consumers use AI assistants to research products, and 31% use them to search for deals.

Since Meta unveiled Muse two weeks ago, the app has amassed more than 2.5 million downloads, according to Sensor Tower data shared with CBS News. By comparison, OpenAI's ChatGPT AI app had 3.1 million global downloads over the same time period, while Anthropopi's Claude tool had 200,000, the market research company found.

In another sign of Muse's growing relevance, Amazon confirmed to CBS News that it has blocked the AI app from shopping on its e-commerce platform, citing security and user experience concerns.

The evolution of search

Until recently, people have mostly used AI as a "glorified search engine," he added. By contrast, Meta is marketing Muse as a tool that will help consumers with everyday tasks and otherwise make their lives easier.

"If [Muse] maintains momentum, it could maybe leapfrog Claude among consumers and get into the top tier" of AI tools, said John Mercer, head of global research at Coresight.

Unlike traditional chatbots, Muse can take action on behalf of users, handling tasks like booking a flight, decluttering people's inboxes and tracking price drops. When it comes to shopping, the benefit for customers isn't just saving time; it's also about helping them find deals, Mercer said.

"It's a natural evolution from search," Mercer said.

In a recent Coresight survey of more than 1,000 consumers, 30% of respondents said they would use an AI tool to compare prices for holiday shopping. The same percentage said they would also use the tools to find deals or promo codes on holiday gifts.

"What agentic commerce does is basically summarizes, discovers, showcases what it thinks is interesting to each individual, and that does cut out some of the work so you don't have to go to multiple sites and cross-compare," Saunders said.

Although Muse is already making a splash, Saunders thinks broader consumer adoption of AI shopping agents will take time. Cases of agentic AI mishaps have already emerged, shaking people's trust in the tools.

"People are still a little bit nervous about allowing AI agents to do the purchasing on their behalf," he said.

Meta's Muse personal agent is a hit, as AI commerce starts to take off

As discussions intensify regarding the emerging realm of artificial intelligence and its implications, experts caution that China is cleverly eroding the United States' global dominance through strategic AI maneuvers, even as President Xi Jinping prepares for his visit to the U.S. this week. The paradigm shift involves a subtle, three-tiered strategy that encompasses extracting knowledge from advanced American AI systems, sabotaging U.S. developers with free open-source alternatives, and ultimately transforming these technologies into tools of cognitive warfare.

Analysts warn that China’s aggressive tactics involve circumventing immense investments made by U.S. research institutions. They exploit advanced American AI frameworks to replicate and enhance their capabilities, all while saturating the market with accessible alternatives. Leah Siskind, an AI research fellow at the Foundation for Defense of Democracies, remarked, “I believe framing this as a ‘race’ is misleading. It resembles a multi-front conflict – we face victories in some arenas but setbacks in others.”

While major American tech companies have invested heavily in building proprietary, high-cost systems, China has taken a markedly different approach by spreading free, open-source AI models worldwide. Siskind cautioned that the financial advantage of utilizing cost-effective Chinese solutions leads some Fortune 500 companies to inadvertently open themselves up to risks such as censorship from the Chinese Communist Party (CCP) and potential backdoor vulnerabilities. “How can one compete with free software when American technology is prohibitively expensive?" she questioned. "Ensuring that a Chinese open-source model is genuinely free of CCP influence is virtually impossible, and that uncertainty is intentional.”

Gordon Chang, a China specialist, highlighted that Beijing is employing these free models as part of a broader economic strategy reminiscent of its earlier maneuvers within the solar panel industry. He believes China aims to undercut American models on cost and availability to ultimately achieve global supremacy in AI.

To successfully flood the market with these free solutions, China first requires access to advanced technologies and a way to overcome the significant lead the U.S. holds in computing infrastructure. Currently, the U.S. boasts over 5,400 data centers compared to around 450 in China and has far superior semiconductor technology. To overcome these challenges, Siskind stated that Chinese research facilities are utilizing a technique known as “knowledge distillation.” This involves smaller models extracting logic and capabilities from larger, more advanced American models, allowing them to make significant strides with minimal investment. 

The implications of this technological rivalry are very real. Recently, the Department of Homeland Security issued a cybersecurity advisory, noting that Chinese firms are running "industrial-scale knowledge distillation campaigns" aimed at harvesting proprietary capabilities from leading American AI systems. 

Furthermore, experts warn of China’s active integration of AI into critical infrastructure and public influence initiatives. Instead of focusing solely on fantastical notions of "killer robots," Siskind emphasized that China's immediate goal is the physical application of AI—integrating it into factory automation, drones, and autonomous logistics systems.

Chang voiced a concerning prospect about AI-fueled propaganda, which poses a significant threat to American democratic processes. “They excel in cognitive warfare and can amplify these capabilities through artificial intelligence. We cannot allow our electoral outcomes to be influenced by China," he warned.

As Xi Jinping’s imminent U.S. visit unfolds, experts advocate for a concerted response from the White House to address these non-military yet highly pertinent threats with the same urgency reserved for conventional military posturing. Siskind urged that to maintain the U.S.’s technological edge, policymakers should close existing loopholes in chip exports, blacklist foreign labs known for deceptive practices, and foster domestic open-source development that aligns with democratic ideals.

Chang underscored the broader implications of this technological struggle, asserting that Xi Jinping's ambitions echo an imperial mindset that could reshape global governance under Chinese hegemony if left unchecked. 

Meanwhile, during a recent address at the United Nations General Assembly, President Donald Trump reaffirmed the U.S. commitment to outpacing China. He announced a shift in terminology, declaring that U.S. government documents would now refer to AI as "superintelligence," emphasizing that the nation remains at the forefront and intends to sustain that lead.

Experts caution that China's actual AI threat to America is already present, surpassing concerns over killer robots.

A moody, dimly lit war room bathed in warm shadow: on one wall a large Chinese flag hangs half-folded, its red fabric catching the last of a single overhead lamp. On a heavy wooden table below, glowing red circuit-board traces snake over a map of the United States, converging on a small, shadowed model of the Capitol building—an unseen AI threat silently closing in.

Beyond killer robots: China's real AI threat to America is already here, experts warn - Fox News

- Experts warn China is undermining U.S. AI dominance through free, open-source software strategies.  
- Beijing's "knowledge distillation" allows copying American AI capabilities, impacting global competition.  
- The U.S. must enhance chip export controls and support domestic open-source AI development.  

NEW You can now listen to Fox News articles!

While Americans brace for a futuristic AI war of killer robots and bioweapons, experts warn ahead of President Xi Jinping's U.S. visit this week that Beijing is already actively undermining America's global supremacy through AI without ever firing a shot.

Experts who spoke with Fox News Digital outlined a stealthy, three-part playbook, which includes extracting the logic of American models, undercutting U.S. developers with free open-source software and ultimately weaponizing those systems for things like cognitive warfare.

They laid out how China bypasses billions of dollars in U.S. research through labs that interrogate cutting-edge American AI and copy its capabilities, while simultaneously flooding the global market with free alternatives.

"I think a ‘race’ is the wrong metaphor for this," Leah Siskind, an AI research fellow at the Foundation for Defense of Democracies, told Fox News Digital. "I think it's kind of like a multi-front war. And in some places, we're winning, and some places we're losing."

XI PREPARES CHINA ‘FOR BATTLE’ AS EXPERT WARNS OF GRAVE DANGER FACING US AHEAD OF DC SUMMIT

While top American tech firms have poured billions into building ultra-expensive, proprietary systems, China has focused heavily on distributing free, open-source models across the globe. Siskind warned that because American tech is costly and the U.S. sector has largely neglected open-source investment, many Fortune 500 companies are making the financial choice to run their operations on Chinese models, unwittingly opening the door to potential CCP censorship, hidden backdoors and, ultimately, global dominance.

"How can you compete with something that's free when American tech is very, very expensive?" Siskind said. "How can we ensure that a Chinese open-source model would be truly clean, that it'd be truly free of CCP censorship? We can't be — we just wouldn't know. And that's by design. We wouldn't know what sort of trick backdoors might be implanted and how that could rear its head later on."

China expert Gordon Chang warned that Beijing is intentionally using these free models to repeat an economic tactic it previously deployed to corner a different industry.

"This is a tactic that China used with solar panels," Chang told Fox News Digital. "Beijing sees that one day it will drive out the American models on cost and availability, and they will then dominate global AI."

CHAD WOLF: CONGRESS HAS A CHANCE TO STOP CHINA FROM TURNING YOUR CAR INTO A CHINESE SPY



But to flood the market with free software, Beijing first needs the underlying technology and a way to bypass America's massive lead in computing power.

Chang noted that the U.S. possesses a commanding lead in physical infrastructure, boasting over 5,400 data centers compared to China's roughly 450, along with far superior microchips. To work around the staggering hardware and energy costs required to train artificial intelligence from scratch, Siskind warned that Chinese labs are closing the gap through a process known as "knowledge distillation," a method where a smaller model essentially interrogates a larger, smarter American model to copy its logic.

"They are able to leapfrog the process by kind of stealing this information," Siskind said, pointing to the Chinese model DeepSeek as a prime example. "This is how they were able to release models like DeepSeek in 2025, which was seismic news because they were able to show that they were really competing with cutting-edge American models and had invested a fraction of the resources."

US ADVERSARY TURNS UP THE HEAT ON AMERICA’S AI LEAD WITH AN UNDERESTIMATED EDGE

The threat is not theoretical. Just last week, the Department of Homeland Security issued a joint cybersecurity advisory warning that Chinese companies are conducting "industrial-scale knowledge distillation campaigns" to extract proprietary capabilities from U.S. frontier models.

Beyond economic undercutting and software theft, the experts also warned Beijing is simultaneously deploying artificial intelligence into physical supply chains and public influence campaigns.

Rather than sci-fi "killer robots," Siskind emphasized that China's immediate physical priority is "embodied AI," embedding artificial intelligence into real-world infrastructure like factory automation, commercial drones and autonomous supply chains.

Chang warned that AI-enhanced propaganda poses a direct threat to American decision-making and democratic sovereignty.

"They're very good at cognitive warfare, and they can enhance their capabilities with artificial intelligence," Chang said. "We do not want our elections determined in China. We do not want our critical national security decisions being made in Beijing. And that's not impossible."

CHINA EXPERT WARNS BEIJING WANTS AMERICANS — NOT THE CCP — TO SPREAD ITS MESSAGE

As President Xi Jinping arrives in the U.S. this week, the experts who spoke to Fox News Digital said the White House must treat these non-cinematic threats with the same urgency as traditional military posturing.

To protect America's remaining edge, Siskind said Washington must tighten chip export loopholes, blacklist foreign labs engaging in deceptive distillation and incentivize U.S. open-source development, emphasizing that "it is very important that the most advanced systems are developed under democracy and represent democratic values, as opposed to authoritarian values."

CHINA-LINKED HACKERS INFILTRATED US GOVERNMENT NETWORKS BEFORE FBI TAKEDOWN

For Chang, the stakes of that technological defense extend far beyond standard economic competition.

"Xi Jinping throughout this century has been pushing the Chinese imperial-era notion of Tianxia, or 'all under heaven' that China's rulers have a right to rule the world," Chang warned. "That sounds ludicrous to us, but a Chinese regime powered by AI might actually be able to accomplish that. We certainly do not want our decisions being made in the Chinese capital."

CLICK HERE TO DOWNLOAD THE FOX NEWS APP

Addressing the United Nations General Assembly on Tuesday, President Donald Trump reiterated Washington's goal of outpacing China, declaring that U.S. government documents will now refer to AI as "superintelligence."

"The United States totally rejects any attempt to construct a globalist scheme to control the artificial intelligence being spoken of," Trump told world leaders. "Whoever wins superintelligence wins. And we're leading now over China by a lot and everyone else. And we're going to keep it that way."

Beyond killer robots: China's real AI threat to America is already here, experts warn

Ulanqab, known for its expansive steppes and volcanic landscapes, is emerging as an unexpected hub for digital innovation. Once celebrated as China’s official "potato capital," the region is now witnessing a transformation as numerous data centers are being built across its farmland. This surge in digital infrastructure coincides with discussions between Donald Trump and his Chinese counterpart Xi Jinping, who are focusing on AI safety, among other critical topics.

China's AI: Data centers emerge from the expansive landscapes of Inner Mongolia

A warm sunrise bathes the rolling grasslands of Inner Mongolia as a sleek cluster of low-profile data centers emerges from the plains. Solar panels shimmer around the metal-clad buildings, and a single Chinese flag waves proudly on a tall pole in the foreground. The scene is lit by soft natural light, blending cutting-edge technology with the vast, open landscape.

China AI: Data centres rise from the rolling plains of Inner Mongolia - BBC

- Ulanqab, known as China's "potato capital," is transforming into a digital hub.  
- The region is developing numerous data centers amidst growing AI safety discussions.  
- Ongoing talks between Donald Trump and Xi Jinping highlight the importance of AI regulation.  

Ulanqab, with its steppes and volcanoes, is an unlikely digital frontier. For years, it has been China's official "potato capital". But now its fields have also become construction sites for a slew of data centres, as Donald Trump and his Chinese counterpart Xi Jinping meet to discuss AI safety among other things.

China AI: Data centres rise from the rolling plains of Inner Mongolia

Laura Bicker from the BBC is reporting live from Inner Mongolia, where multiple data centers are rapidly being constructed as China seeks to become a frontrunner in artificial intelligence on a global scale.

Bicker highlights the contrasting perspectives that Chinese and American citizens hold regarding the data center sector, particularly in light of the upcoming meeting between leaders Xi Jinping and Donald Trump in Washington later this week. 

For deeper insights into this evolving story, click here.

Uncommon glimpse at the distant data centers driving China's AI goals

A dawn-lit plateau deep in western China, where sleek modular data-center buildings stretch across rocky terrain under soft golden sunlight. A single Chinese flag flutters atop the nearest unit as mist coils around low cooling towers. A lone technician in a hard hat and jacket walks along gravel paths toward the entrance, rugged mountains rising behind.

Rare look at the remote engine rooms powering China's AI ambitions - BBC

- China's construction of data centres aims to dominate the global artificial intelligence sector.  
- Attitudes towards data centres vary significantly between China and the United States.  
- The article coincides with an upcoming meeting between Xi Jinping and Donald Trump.  

The BBC's Laura Bicker is in Inner Mongolia where dozens of Chinese data centres are being built, as the country aims to lead the world in artificial intelligence.

She explains how attitudes differ between people in China and the US when it comes to the data centre industry, ahead of leaders Xi Jinping and Donald Trump's meeting in Washington later this week.

Read more on this story here.

Rare look at the remote data centres powering China's AI ambitions

On Tuesday, Anthropic unveiled its latest model, Opus 5.5, marking a significant advancement in performance for coding and cognitive tasks, the company announced. As the highest-performing and most expensive variant in Anthropic's Claude series, Opus 5.5 stands out, with Sonnet offering mid-tier capabilities and Haiku being the most budget-friendly and fastest option. Remarkably, Anthropic claims that this new model surpasses the larger Fable model in numerous benchmarks and successfully executes informal tasks where Fable struggled.

In addition to its superior performance, Opus 5.5 comes at a lower cost than its predecessor. The rate for output tokens is set at $20 per million tokens, down from the previous $25. Similar reductions in pricing are observed across various metrics, and the model operates more swiftly due to a decrease in the computational power required for its operation.

Opus 5.5 also introduces notable improvements in communication style, employing less jargon and prioritizing critical information at the beginning of its responses.

This launch follows closely on the heels of Opus 5, which debuted on July 24. Anthropic has indicated that enhancements for Sonnet 5.5 and Haiku 5.5, the succeeding models in the series, are forthcoming within the next few weeks.

Anthropic noted that Opus 5.5 shares comparable capabilities in biology and cybersecurity with the Mythos model, meaning it will also adhere to the same safety protocols as the Fable model. These safeguards restrict the model's functionality regarding the exploration of vulnerabilities in compiled programs or the development of recognizable biological weapons, among other applications.

The release of Opus 5.5 is significant as it marks the first model introduced since CEO Dario Amodei advocated for a more measured approach to advancing AI technologies. He expressed the necessity of aligning the pace of AI advancement with efforts to mitigate risks effectively. "I have become convinced that fully addressing the risks requires even more prudence," Amodei wrote in a recent post, emphasizing the importance of ensuring that advancements in risk prevention keep pace with improvements in capabilities.

Opus 5.5's safety training mirrors that of its predecessors, featuring alignment testing and evaluations conducted by external organizations such as METR and Frontier Design. However, Anthropic highlighted the development of more sophisticated training and evaluation processes for future models, including enhanced security and monitoring systems.

"As AI continues to evolve, it's essential that public policy increasingly ensures the safety of systems people depend on," the blog post stated. "Building this capacity takes time, and we have begun laying the groundwork to facilitate it." The company plans to share further details about these initiatives in the near future.

Anthropic launches Opus 5.5 featuring reduced prices and Fable-equivalent performance.

A high-tech command center bathed in warm morning sunlight, with a prominent Anthropic logo embossed on a sleek console in the foreground; a large screen reads “Opus 5.5,” beside a golden quill soaring above a dragon-shaped performance graph climbing skyward; falling price-tag icons below symbolize the new lower costs.

Anthropic releases Opus 5.5 with lower prices and Fable-level performance - techcrunch.com

- Anthropic launched Opus 5.5, outperforming Fable in coding and knowledge work benchmarks.  
- Opus 5.5 is cheaper and faster, charging $20 per million tokens versus $25 previously.  
- The model implements new communication strategies, prioritizing clarity and important information delivery.  

Anthropic’s newest model, Opus 5.5, was released on Tuesday, setting a new state-of-the-art in coding and knowledge work performance, according to the company. Opus is the most capable and expensive tier in Anthropic’s three-tier Claude lineup; Sonnet sits in the middle, while Haiku is the fastest and cheapest. Notably, Anthropic says, the release outpaces the larger Fable model in many benchmarks and succeeded in a number of informal tasks that Fable failed to complete.

The new model is also significantly cheaper than its predecessor. Output tokens will be charged at $20 per million tokens for Opus 5.5, compared to $25 for the previous model. Other metrics have similar price drops. The model is also faster to run, reflecting an overall drop in the compute required to serve it.

The new version also makes significant changes to how Opus communicates, with the Opus 5.5 less likely to use jargon and more likely to put important information at the start of its messages.

The launch comes just two months after the release of Opus 5 on July 24. According to the announcement, Sonnet 5.5 and Haiku 5.5, which are the next tiers in the lineup, will be released “in the coming weeks,” with similar performance improvements.

Anthropic says that Opus 5.5 is comparable to Mythos in its biology and cybersecurity capabilities, so its release is subject to the same safeguards as the company’s Fable model. Those safeguards limit how much the models can be used to discover exploits in compiled programs or developing recognizable biological weapons, among other tasks.

Opus 5.5 is Anthropic’s first model release since CEO Dario Amodei embraced calls to pace the frontier, deliberately slowing down progress on AI capabilities to match the rate of progress on alignment.

“I have become convinced that fully addressing the risks requires even more prudence,” Amodei wrote in a post earlier this month, “not just investing in risk prevention, but pacing the rate of capabilities advancement so that risk prevention has time to keep up.”

Opus 5.5’s safety training was broadly similar to its predecessors, with alignment testing and pre-release evaluation by outside organizations like METR and Frontier Design. But Anthropic emphasized that more advanced training and evaluation systems were already being prepared for future models, including improved security and monitoring systems.

“As AI becomes more capable, public policy should play a larger role in making sure the systems people rely on are safe. That capacity takes time to build, and we’ve started to put the infrastructure in place to support it,” the blog post reads. “We expect to share more details on these efforts soon.”

Basecamp Research: We require AI that naturally comprehends biology

More from AI

We're expanding AI Brief to additional languages and introducing a new AI Max reporting feature.

OECD anticipates that the AI surge will alleviate the impact of the energy crisis in the Middle East for the time being.

Muse AI Elevates Expectations for Meta Connect Keynote

Your options for privacy