Google says AI-powered hacking has reached industrial scale.
What happened: Google’s threat team says criminals used AI to help develop a working zero-day exploit that could bypass two-factor authentication, pushing cyber risk from theory into live operational reality.
Why it matters: The operational risk is no longer just more phishing and spam. It is that AI can speed real exploit development and reduce the time between model capability and criminal use.
SoftBank’s France talks showed the AI infrastructure race is now partly a national power-system race.
Bloomberg reporting said SoftBank was weighing a major French AI data-center project, with France’s nuclear-heavy grid presented as a strategic advantage.
A handheld AI microscope pushed cancer detection closer to the point of care.
Rice and MD Anderson researchers said their new device could deliver high-resolution, real-time imaging without the usual tradeoff between detail and coverage.
OpenAI’s enterprise deployment move hit India’s IT-services trade.
Reuters-linked market coverage said India’s IT shares fell toward a three-year low as investors worried OpenAI’s enterprise deployment venture could pressure traditional outsourcing and consulting models.
Workers overwhelmingly support AI guardrails that keep humans in the loop.
A new AFL-CIO poll found very broad support for human final decision-makers, disclosure of AI monitoring, and stronger worker protections as automation expands into workplace management.
Tech job postings hit a three-year high as AI demand fuels hiring.
The April hiring data pointed to a real rebound in tech openings, especially in roles that support the AI stack. The jobs picture was getting more uneven, not simply worse.
SoftBank’s France talks made the AI infrastructure race look like energy policy.
Bloomberg reporting said SoftBank was weighing a massive French AI data-center project. Macron’s pitch was that France could offer low-carbon nuclear power at national scale.
Pennsylvania was building an actual chatbot enforcement playbook.
A state task force was looking for AI chatbots that misled users into believing they were licensed professionals, after the state sued Character.AI over an alleged fake doctor persona.
The public trail around frontier-model safety testing just got harder to follow.
Reporting on the Commerce Department removing details of a model-testing agreement underscored a basic policy problem: AI oversight cannot be credible if public documentation disappears when scrutiny rises.
A handheld AI microscope aimed to spot cancer earlier at the point of care.
Rice and MD Anderson said PrecisionView could capture high-resolution, real-time tissue imaging without forcing clinicians to trade detail for coverage.
Cleveland Clinic’s GenT framework tried to make genetic signals more actionable for drug discovery.
The AI method grouped variants around genes instead of treating millions of variants one by one, a shift designed to surface more useful drug targets in complex neurological disease.
AI-enhanced video made a real shooting story harder to read accurately online.
Poynter showed how AI “enhancement” of low-quality footage from the White House Correspondents’ dinner shooting introduced distortions that looked more authoritative.
The K-12 debate was shifting toward pause-the-rollout politics.
Some educators and advocates wanted moratoriums on student-facing generative AI while districts sorted out privacy, developmental, and equity concerns.
