ai risk today logo svg bl font
Menu
SUBSCRIBE
SUBSCRIBE
ai risk today logo svg bl font
Menu

AI risk 2026

Abstract blue wireframe cube structures with golden cores depicting AI building itself, for coverage of Anthropic's call for a global AI pause.

Anthropic Calls for Global AI Pause, Says AI Nearly Builds Itself

, , , , , , , , , , , ,

The company’s research arm says Claude authors more than 80% of its production code and warns that AI designed to build its own successors could arrive before governments can respond.

 

The Anthropic Institute published an analysis this week showing that AI is already accelerating its own development and called on governments, rival labs, and civil society to build the systems needed for a global AI pause. Anthropic said it would stop if other leading labs agree to pause development, as long as there is a way to verify everyone adheres to the agreement.

The report uses publicly available benchmarks alongside previously undisclosed internal Anthropic data.

The central finding: AI is no longer just a tool used in development. It is now a driver of it.

 

AI already does most of the work at Anthropic

Claude now writes more than 80% of the code that goes into Anthropic’s systems. Engineers are now reviewers rather than writers. That shift increased the amount of code each engineer ships per day by 8x since 2024, and employees say they get around 4x as much work done.

The range of tasks AI can handle without human help doubles every four months. In March 2024, Claude could complete tasks a human finishes in four minutes. By April 2026, it was handling work that takes a skilled person 12 hours.

On open-ended research tasks, Claude’s success rate reached 76% in May 2026, up 50 percentage points from six months earlier. On deciding what to research next, Claude now outperforms human researchers 64% of the time in live sessions, up from 51% in November 2025.

 

Once AI builds itself, controls become much harder to add

The analysis lays out three possible futures. In the first, AI capability growth stalls and today’s AI spreads gradually through the economy. In the second, AI handles more and more of the work while humans still set the direction. In the third, AI systems design and train their own successors without meaningful human input. Researchers call this recursive self-improvement. Anthropic says the first scenario is unlikely.

If AI systems can design their own successors, the pace of progress is no longer limited by how fast humans can conduct research. It becomes limited only by computing power. At that point, the controls needed to keep AI systems safe become much harder to build after the fact.

 

A pause only works if everyone can prove they stopped

Anthropic said that an individual company can easily pause its development, but that would only change who leads the race, not alter the pace of development. A real coordinated pause requires a way to verify that all frontier AI developers stop at the same time. The report points out that AI training runs are far easier to hide than missile silos, and that comparable arms control treaties took decades to build.

Anthropic says it will organize conversations with policymakers, researchers, civil society, and other AI companies in the coming months and will publish the results.

Abstract visualization of a dark cracked sphere with explosive blue and gold rays representing Anthropic's Project Glasswing expansion to critical infrastructure sectors.

Anthropic Expands Project Glasswing to 150 New Organizations Across 15 Countries

, , , , , , , , , , , ,

Power, water, and healthcare companies join the program as Anthropic warns that within a year, AI models will become more powerful, cheaper, and widely available, including to cybercriminals. 

 

Anthropic is extending Project Glasswing to approximately 150 new organizations in more than 15 countries, adding sectors not represented in its initial cohort, including power, water, healthcare, communications, and hardware. Each new partner must meet Anthropic’s security requirements before gaining access to Claude Mythos Preview (“Mythos”).

In April, Anthropic announced that roughly 50 initial partners, including NVIDIA, Apple, Google, Microsoft, and Amazon, had begun using Mythos to scan their codebases. Those partners found more than 10,000 high- or critical-severity security flaws in the first weeks of the program.

The new partners share one characteristic: scale of exposure. For most, Anthropic estimates a successful attack on their codebase could affect more than 100 million people, with consequences for both national and global security. Many are vendors whose software underpins systems relied on by other organizations worldwide, including governments.

Anthropic increased participation out of concern that other AI companies will likely have Mythos-class models within 6 to 12 months and could release them without safeguards that prevent misuse. In that scenario, AI-assisted cyberattacks become significantly more common and harder to predict. The company frames Project Glasswing as a way to establish operating norms before that happens.

Anthropic notes that finding vulnerabilities is no longer the bottleneck: getting them patched is. Partners now use Mythos to write patches and run pre-release checks on their own code, not just to surface problems. Anthropic says the model can also be used for penetration testing, threat detection and response, and rebuilding legacy codebases in memory-safe languages. This week, Google released AI Threat Defense, an advanced cyber detection model that also automatically applies patches.

Anthropic said it plans to further expand Glasswing, prioritizing critical infrastructure providers, open-source software maintainers, and safety testers. It is also scaling its Cyber Verification Program, which would grant Mythos-class capabilities to more organizations for specific defensive tasks. General access to Mythos, the company said, requires safeguards no AI developer has yet built.

Silhouetted crowd before a glowing futuristic city with digital data patterns, representing public skepticism and demand for AI regulation.

Americans Sour on AI; 65% Across Party Lines Want More Regulation

, , , , , , , , ,

Only 17% expect AI to have a positive impact over the next decade, and even among optimists, 43% say the government has done too little.

Key Takeaways

  • Only 17% of Americans expect AI to have a net positive impact on the U.S. over the next 10 years; 42% expect it to be negative
  • 65% say the government has done too little to regulate AI, including 77% of Democrats, 72% of independents, and 53% of Republicans
  • Even among Americans who expect AI’s impact to be “very positive,” 43% say the government has done too little
  • 41% of employed Americans worry about losing their job or having their hours cut due to AI

The Annenberg Public Policy Center of the University of Pennsylvania released new national survey data this month. The survey asked 1,330 U.S. adults about their views on AI and what they think the government should do about it. The picture that emerges is skepticism, not enthusiasm.

 

Most people expect AI to have a negative impact

Only 17% of respondents expect AI to have a somewhat or very positive impact on the United States over the next 10 years. 42% expect the impact to be negative. About a third say the effects will be roughly balanced.

The pessimism is not coming from people who haven’t engaged with the technology. 78% say they’ve heard at least a moderate amount about AI. 67% report using it at least a few times in the past month.

 

Bipartisan demand for AI regulation

65% of Americans say the government has done too little to regulate AI. The breakdown by party: 77% of Democrats, 72% of independents, and 53% of Republicans. When asked which level of government should lead, 52% chose the federal government.

The demand for legislation intensifies with pessimism, but isn’t confined to it. Among respondents who expect a “very negative” impact from AI, 83% say the government has done too little. Among those who expect a “very positive” impact, the number is still a significant 43%.

 

Workers across party lines are concerned about their jobs

41% of employed respondents say they worry about losing their job or having their hours reduced because of AI. Democrats (50%) express more concern than independents (41%) and Republicans (32%).

The APPC fielded the survey Feb. 17–March 20, 2026. Several states have recently accelerated legislation, including Colorado, which ratified bill SB-189, and Missouri, which introduced an eight-bill legislative package.