AI, security and power
Cybersecurity, biosecurity, regulation and infrastructure: the decisions shaping how AI is deployed and who carries the risk.
What has been approved, who can enforce it and how it may affect you without turning every announcement into a dystopia.READING ROUTE
THE DOSSIER
Start here and follow the story.
01OPENOpenAI says Astra is coming soon, its first model rated Critical for cybersecurity
Hours after Claude Fable 5.1 arrived, OpenAI revealed that Astra found two zero-days, escaped a hardened browser sandbox and reached root on a protected operating system. The model is nearing release, but its strongest cyber abilities will be restricted.
02OPENOpenAI halts its biggest AI training run over what its next model can already do
Astra may hit OpenAI's critical cyber threshold, so the company stopped its largest training run. Altman says they will act alone if they must.
03OPENOpenAI disbands the team guarding against its gravest risks
The tests move into the biosecurity and cybersecurity teams. Integration on paper; in practice, no one holds the authority to stop a dangerous model.
04OPENNVIDIA brings together 40 giants to build an army against AI-powered attacks
Microsoft, Cloudflare, CrowdStrike, IBM, Cisco and Hugging Face are among the founding members of the Open Secure AI Alliance, which aims to build open defenses against malicious AI agents.
05OPENEurope forces Google to open Android: ChatGPT, Claude or Perplexity will be able to truly replace Gemini
Rival assistants will be able to wake by voice, understand what is on screen, use apps and change phone settings. It will not happen immediately, but the European Union has just struck at Gemini's biggest advantage: being built into Android.
06OPENEurope fines Google €890 million. The number is the least of it
Brussels sanctions Google for favoring its own services in Search and for making it hard for developers to point users to cheaper purchases outside Google Play. It has 60 days to change.
07OPENOpenAI and Anthropic want to slow Chinese AI. Jensen Huang thinks they are digging their own grave
Silicon Valley's two fiercest rivals are aligning in Washington against Chinese open-weight models. NVIDIA and many startups think shutting them out could hand China the very victory America wants to prevent.
08OPENTrump jokes about a fourth term at the correspondents' dinner. Behind the gag sits a very real AI policy
At the toned-down edition of the event, rescheduled after April's attack, the president needled the press with the fourth-term line. What interests us is the rest: his administration is the one deciding which AI models ship and which stay in the drawer.
09OPENChatGPT has been failing across half the world all morning. OpenAI's status page insists everything is fine
The outage started at 4:47 AM Eastern, with logins failing, chat histories vanishing and API requests bouncing. OpenAI briefly acknowledged the incident on its status page, which went back to green while thousands of users were still locked out
10OPENThe AI chip war is a fight against NVIDIA's bill
Amazon, Google, Microsoft and Meta are building chips to lower the cost of their own AI. The surprising part is that they are still paying NVIDIA millions.
11OPENHow to tell if an image or text was made by AI: the tricks that work and the myths that don't
Counting fingers no longer cuts it and text detectors lie more than they hit. We explain which signs truly give away an AI image or text, where to check the invisible labels Europe now requires, and why your best tool is still to distrust the context.
12OPENAs of today, AI has to confess in Europe: what must be labelled and what escapes until 2027
The EU AI Act enters its most talked-about phase today, 2 August: chatbots must warn they're machines and deepfakes must be marked as synthetic. But the technical watermark has a reprieve until December, and the biggest part, high-risk systems, doesn't arrive until 2027.
12 of 27 stories showing