Weeks ago Jensen Huang said labs should pace themselves if they felt out of control. Now he puts the risk at 0%.
THE SO WHAT
When the leading GPU supplier says there is a 0% chance AI ends the world by 2030 and urges the industry to go "as fast as it can," he is giving political cover for maximal deployment and capex. If you’re an operator, assume less regulatory drag than the doomer narrative implies and plan for faster model iteration, but do your own risk thresholds instead of outsourcing them to suppliers.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIA look at AI safety groups METR, Redwood Research, and Apollo Research, as AI misalignment incidents at OpenAI and Anthropic thrust them into the spotlight
Independent groups like METR, Redwood, and Apollo moving from the margins into the spotlight after misalignment incidents means third-party evals are becoming part of the de facto governance stack. If you’re deploying frontier models, expect board and regulators to ask which external evaluators you use—not just what your internal red-teaming looks like.
Applied AIChina State TV Affiliate Flags Anthropic Data and Privacy Risks
A China Central Television–affiliated account framing Anthropic’s revised privacy policy as enabling data sharing with US intelligence—without legal process—shows how AI vendor policies are now fodder for geopolitical narratives. If you operate in or near China, assume your AI stack choices will be scrutinized as national-security decisions, not just IT procurement.
Applied AIThree researchers used Claude to reach OpenAI’s internal code. OpenAI paid $6,500 and closed the hole in 14 hours.
Researchers chaining two weaknesses and using Claude to reach OpenAI employee accounts and internal code in under 72 hours—then getting $6,500—shows AI-augmented offense is cheap while the potential blast radius is huge. If you run bug bounties or internal red teams, assume attackers have LLM copilots and raise both rewards and response readiness accordingly.
Applied AIAI-Generated ‘Actress’ Tilly Norwood Starts Rambling in Cantonese During Piers Morgan Interview
Synthetic influencers are moving from novelty to reputational risk surface—one glitchy interview can reset brand and audience trust. If you're experimenting with AI personas, treat them like live broadcast systems: preflight, guardrail, and have a human failover plan.