Anthropic warns of ‘existential risks to humanity’ from AI; AstraZeneca makes $2bn cancer drug tie-up – business live
THE SO WHAT
Anthropic putting ‘existential risk to humanity’ language directly into IPO materials moves AI safety from blog posts into securities disclosures. Boards now have cover—and pressure—to treat frontier AI risk as a formal governance topic, not a side conversation between CTO and research leads.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIOpenAI cancels October launch of GPT-6.1 Astra after failed safety tests
A pulled GPT-6.1 Astra release after failing internal safety and alignment bars shows that frontier cadence is now gated as much by governance as by capability. If your roadmap assumes predictable version bumps from any single lab, build in vendor diversification and feature flags so you can absorb slips like this without rewriting plans.
Applied AIMeta signs deal for Firmus AI computing capacity in Southeast Asia
Meta renting dedicated AI chips from Firmus in Southeast Asia underscores that hyperscalers are now multi-sourcing compute like a strategic commodity. If you’re a heavy AI user, start treating regional compute diversity—different vendors, different jurisdictions—as a resilience and latency decision, not just a cost line.
Applied AIOpenAI Scraps Debut of Latest Astra Model
OpenAI shelving GPT-6.1 Astra despite improvements on ‘model laziness’ highlights a new tradeoff surface—capability gains that don’t clear safety evals won’t ship. For teams building on top of these models, assume behavior can change materially between internal previews and public release, and keep regression tests close to your critical workflows.
Applied AIAnthropic’s prospectus details losses, growth, and, yes, a warning that its AI could end humanity
Anthropic telling investors it’s losing tens of billions annually, growing fast, and might be building tech that could ‘end humanity’ crystallizes the new AI capital compact—huge burn, huge upside, explicit tail risk. If you’re raising in this space, expect LPs and boards to ask not just about runway and TAM, but about your safety posture and how you’d respond if your own risk estimates worsened.