Top Stories
All Sections
Finance
About
16 results for "AI safety"
For the first time, the CEOs of OpenAI and Anthropic addressed the UN Security Council together, warning that self-improving AI systems could slip beyond human control.
Anthropic disclosed that Claude now leads approximately 26% of the company's own model research and development — roughly one in four R&D cycles initiated by the AI itself.
King Charles III convened the CEOs of OpenAI and Anthropic and co-founders of Nvidia and Google DeepMind at Dumfries House, warning that AI risks developing 'darker capacities — perhaps even to take life.'
The three dominant AI labs have been in private multi-week negotiations on coordinating frontier development pacing — framed as safety, but with significant cash conservation implications.
Anthropic disclosed blocking queries that could have aided bioweapon development and uncovering a coordinated AI-agent campaign that breached 395 organizations — one of the most specific public admissions of AI weaponization at scale.
California's Safe and Secure Innovation for Frontier AI Models Act passed both chambers and awaits Governor Newsom's signature by September 30, which would make California the first US state to mandate pre-deployment safety testing for large AI models.
Anthropic is developing AI agents that adapt based on real-world performance feedback — a self-improvement capability that raises immediate alignment questions and sets up a race dynamic with OpenAI and Google.
California's Senate voted 39-0 to ban generative AI companion chatbots in toys for children 12 and under, sending SB 867 to the Assembly with the nation's strongest child AI safety signal yet.
Anthropic submitted a confidential S-1 to the SEC on June 1, one week after a $65 billion Series H valued the AI safety company at $965 billion.
Microsoft, xAI, and other major AI companies have agreed to provide U.S. government regulators early access to AI models before public release — the most substantive voluntary federal AI oversight commitment in the U.S. to date.
Tesla launched the Cybercab as a fully driverless commercial robotaxi—but at least 17 accidents involving its autonomous systems since 2025 are drawing regulator scrutiny.
Connecticut's SB5 cleared both legislative chambers on May 1, 2026, establishing AI obligations around companion chatbots, automated hiring, and synthetic content — making it one of the most comprehensive state AI laws in the country.
The Pentagon signed AI access agreements with seven tech firms — and excluded Anthropic after the company demanded safety guardrails on military use. The split lays bare the tension between AI safety commitments and defense contracts.
Anthropic's most capable model ever built, Claude Mythos, will not be publicly released. Fifty organizations get gated access under Project Glasswing to find their own vulnerabilities before adversaries can exploit the model.
Anthropic announced Claude Mythos on April 7 — then withheld it after the model autonomously identified thousands of zero-day vulnerabilities across every major OS and browser.
Anthropic will not release its Claude Mythos model to the general public after internal testing found it could autonomously discover and exploit thousands of previously unknown software vulnerabilities.