- A letter from Apple to U.S. senators revealed Apple privately threatened to pull xAI's Grok from the App Store in January after finding policy violations tied to sexualized deepfakes.
- Apple rejected an initial moderation fix before approving a revised submission, while NBC News reports similar content is still being generated via prompt workarounds.
Snapshot — April 15, 2026
5 stories
- OpenAI Launches GPT-5.4-Cyber — A Frontier Model Built for Defense OpenAI unveiled GPT-5.4-Cyber, a fine-tuned variant of GPT-5.4 specifically optimized for defensive cybersecurity work, with deliberately relaxed guardrails for security-relevant tasks.
- The model is being rolled out on a restricted basis to vetted vendors, researchers, and government teams through an expanded Trusted Access for Cyber (TAC) program.
- Berkeley Researchers Break Every Major AI Agent Benchmark — Without Solving a Single Task Researchers at UC Berkeley's Center for Responsible, Decentralized Intelligence — including Dawn Song, Koushik Sen, and Alvin Cheung — published a paper demonstrating that all eight of the most prominent AI agent benchmarks (SWE-bench, WebArena, OSWorld, GAIA, Terminal-Bench, FieldWorkArena, CAR-bench, and one other) can be exploited to achieve near-perfect scores without actually completing any task.
- Stanford's HAI released its annual AI Index for 2026, finding that AI systems are advancing rapidly in reasoning, coding, and scientific applications — yet public anxiety about AI's effects on employment and society is intensifying in parallel.
- The report highlights a widening trust gap: while enterprise and government adoption is accelerating, public confidence has not kept pace with capability gains.
The Hacker News · Reuters · The Star · MSN · Forbes · VentureBeat · CNBC · TechCrunch · 9to5Google · Insider Monkey · Stanford HAI · UC Berkeley RDI · MIT CSAIL · The Neuron · Bloomberg Law · GizChina · Tech Xplore · The Atlantic · All About Lawyer · HuMai Blog · Axios · MIT Technology Review · OpenAI Newsroom · Tech-Insider