A coalition of state attorneys general is investigating OpenAI, with subpoenas reportedly focused on advertising, data handling, minors, health data, model behavior, and safety policies. For AI operators, the probe is another signal that consumer AI governance and retention practices are becoming board-level compliance issues.
Tom's Hardware reports that US officials warned Anthropic about a jailbreak affecting Fable 5 before export controls took the model offline. The dispute highlights how model security, geopolitical access controls, and benchmark leadership are now tightly linked.
Internet services in Azad Jammu and Kashmir have remained suspended for an eighth consecutive day, forcing residents to travel outside the region for basic online access. The outage is hitting students, remote workers, and freelancers who depend on connectivity for exams, client work, and digital payments.
The federal government plans to modernize the Central Directorate of National Savings through digital payments and service access. The move could reduce branch dependence for millions of savers and push another public-sector finance workflow toward online delivery.
npm 12 will change install behavior so dependency scripts no longer execute automatically unless they are explicitly allowed. This is a major supply-chain hardening step for JavaScript projects because malicious postinstall hooks have been a recurring attack path.
A former Iowa school district IT employee received a 21-month prison sentence after cyberattacks that disrupted classroom operations, deleted accounts, and caused financial damage. The case is a reminder to revoke privileged access quickly and audit admin activity after staff exits.
Phoenix LiveView 1.2 is out, giving Elixir teams a new stable release for building interactive server-rendered apps. The release is worth tracking for teams using LiveView in production because framework updates often affect component patterns, client hooks, and upgrade planning.
KPMG withdrew a report on AI usage after apparent hallucinations were found in the material. The incident underlines the operational risk of publishing AI-assisted analysis without source validation, especially for consulting and compliance-sensitive work.