“Frontier AI” refers to the most advanced, large-scale AI models — the systems pushing the current edge of capability and often at the center of safety and governance debates. This hub covers frontier model releases, capability and risk evaluations, and the safety measures proposed by companies and governments around the world’s most capable AI systems.
The White House told AI firms it finished a voluntary framework for reviewing frontier models' cyber risks on schedule, but is withholding the document itself from the public.
Read more →President Trump's AI executive order set an August 1 deadline for a frontier-model review framework and cyber benchmarks; agencies published nothing by the cutoff.
Read more →
Frontier AI is the industry's term for the most capable AI systems in existence — but there's no single legal definition. Here's how the label actually works.
Read more →More than 1,200 employees at OpenAI, Anthropic, Google DeepMind, and Meta have signed a letter asking Washington to back international tools for pacing frontier AI development.
Read more →
Ilya Sutskever's secretive AI lab has no product, no revenue, and a $32 billion valuation. Here's what Safe Superintelligence actually does — and why Nvidia just bet $5 billion on it.
Read more →Nvidia is investing in Ilya Sutskever's Safe Superintelligence, giving the secretive AI safety lab early access to its next-generation Vera Rubin chips and a tenfold jump in compute.
Read more →
Claude Opus 5 is Anthropic's new mid-tier AI model — nearly matching flagship Fable 5's performance at half the token price. Here's what it is and what it costs.
Read more →Anthropic released Claude Opus 5 on July 24, pairing near-flagship coding and agent performance with Claude Fable 5's results at roughly half the cost, plus gains in science and finance tasks.
Read more →
Claude Fable 5 is Anthropic's most capable publicly available AI model. Here's what it is, how it differs from Claude Mythos 5, and what it costs to use.
Read more →Anthropic mathematician Levent Alpöge used Claude Fable 5 to find a counterexample to the 1939 Jacobian conjecture, and mathematicians including Terence Tao independently verified it within days.
Read more →
The US and China will hold their first AI dialogue under the Trump administration in September, led by Treasury Secretary Scott Bessent, Reuters reports, as both sides look to manage frontier-model risks.
Read more →The Trump administration is increasingly deciding which firms can access frontier AI models from Anthropic and OpenAI, CNBC reports, shifting gatekeeping from the labs to the government itself.
Read more →Demis Hassabis called for a US-led, industry-funded body to test frontier AI models before release, arguing AGI is now only a few years away.
Read more →
A government body that stress-tests powerful AI models for dangerous capabilities before they ship — the closest thing frontier AI has to an independent inspector.
Read more →Claude Mythos 5 is Anthropic's cybersecurity-focused AI model, built to hunt software vulnerabilities — and available to only about 100 vetted partners, not the public.
Read more →A responsible scaling policy is a frontier AI lab's own rulebook for when a model gets too capable to release without extra safeguards. Here is how the idea works and who uses it.
Read more →Ukraine's Ministry of Digital Transformation says it will prioritize AI systems it can run on its own servers, citing lessons from recent US restrictions on Anthropic.
Read more →Z.ai has released GLM-5.2, a 744-billion-parameter open-weight model that matches GPT-5.5 on key benchmarks and trails Claude Opus 4.8 narrowly — trained entirely on Huawei silicon with no Nvidia hardware.
Read more →Meta's superintelligence chief told employees that the company's next model, codenamed Watermelon, has caught up to OpenAI's GPT-5.5 on key benchmarks — though the specific tests and results have not been disclosed.
Read more →Anthropic restored global access to Claude Fable 5 on July 1 after the US Commerce Department lifted export controls imposed 19 days earlier, following a jailbreak discovered by Amazon researchers.
Read more →Chinese technology company Meituan has released LongCat-2.0, a 1.6-trillion-parameter coding model trained entirely on domestic Chinese chips, under the MIT License — the first frontier-class open-source model built without Western semiconductors.
Read more →