Anthropic Researchers Warn AI Could Kill Humans as Cognition Hits $48B Valuation and Suno Strikes Label Deals
A senior Anthropic safety researcher warned there is more than a 10 percent chance artificial intelligence could kill all humans by the end of the decade, according to a post shared internally and reported by multiple outlets. The warning came just hours after a colleague resigned over fears that Anthropic and its rivals are carelessly racing to build superhuman systems they cannot control, marking one of the most public internal fractures at the AI lab backed by billions from Amazon and Google.
Cognition, the maker of the Devin AI coding agent, hit a $48 billion valuation in its latest funding round, signaling investors believe AI coding is far from a winner-take-all market, according to reporting on the deal. The valuation multiple is reportedly higher than Cursor's was before its sale to SpaceX, underscoring how aggressively venture capital is chasing developer tools even as competition intensifies.
Suno rolled out v6, a new line of AI music models trained in partnership with Warner Music Group and BMG, and says it will pay labels and publishers royalties when the models are used, per Bloomberg. The new model was trained on music from Warner Music Group, though users will not be able to reference specific artists in prompts, representing a notable shift toward licensed training data in the generative music space.
Microsoft's October patch release is an unusually heavy one, with security teams pumping out fixes ahead of an expected onslaught of AI-assisted attacks, according to coverage of the release. The rollout addresses dozens of vulnerabilities across Windows, Office, and Azure infrastructure as researchers warn that threat actors are increasingly leveraging large language models to automate exploit discovery.
Hackers are stealing Claude tokens from Anthropic subscribers, with at least one user noticing last month that his account was consuming tokens even though he was not actively working, according to reports. Anthropic has since warned users about the attacks, which appear to target API keys and session credentials to siphon paid usage from Claude accounts.
Meta dragged its feet removing advertisements that pushed apps designed to nudify real teenagers' Instagram photos, with some ads running for weeks before takedown, according to reporting on the investigation. The ads used the tagline "this is the AI men actually use" and targeted imagery of young girls, raising fresh questions about Meta's automated ad review systems and their failure to catch exploitative AI-generated content.
Sources: