AI News Updated 05/012/2023 Meta announces generative AI features for advertisers (techcrunch.com) IBM’s CEO says its new A.I. tools will be able to do ’30 – 50
Apple Intelligence already supports a broad set of languages, while the new Siri AI is expanding beyond English. iPhone Duo shows how Apple is redesigning hardware for personal AI.
Read more →
Apple is tightening macOS Full Disk Access as AI agents gain the ability to read files, operate apps and work autonomously. Permissions, auditing and liability are becoming the next platform battle.
Read more →
OpenAI Dots, Meta Muse, Grok Bot and Manus 2.0 are turning chatbots into persistent cloud coworkers. Here is how their tools, permissions, strengths and limits compare.
Read more →
The AI agent race has changed categories. The question is no longer only which model gives the smartest answer. It is which agent can remember a long-term goal, sign in to real tools, keep working on its own computer and reliably stop when it crosses a boundary.
Read more → AI Agent 的竞争已经换了赛道。过去比的是谁回答得更聪明，现在比的是谁能记住长期目标、登录真实工具、在云电脑里持续工作，并且在该停下时真的停下。
Read more →
The biggest AI story of the week is not a new benchmark. It is a fight over what the technology should be called, who gets to control it, and whether the companies building increasingly autonomous agents can safely police themselves.
Read more →
Claude Opus 5.5 lowers token prices and posts strong independent results, but external testing describes an efficient, incremental upgrade rather than an autonomous-research leap.
Read more →
OpenAI’s independent mathematics advisory group can shape how AI-generated results are reviewed and released, but its role is advisory and the company’s capability claims remain unverified.
Read more →
EPB has combined an installed IonQ quantum computer with its commercial quantum network, creating a practical test bed whose value must now be demonstrated through reproducible workloads.
Read more →
Google and the Gates Foundation are backing a multi-layer agriculture AI program whose success depends on local data, language access, human delivery, and measurable farmer outcomes.
Read more →
OpenAI’s new reporting framework turns six model-misalignment cases into a test of how transparently frontier labs disclose and correct agent failures.
Read more →
Perplexity’s Windows local agent keeps models, files, and workflows on a powerful PC, but its 24GB VRAM floor sharply limits who can run it.
Read more →
Anthropic’s advisor plugin connects Claude to major wealth platforms while keeping regulated decisions and critical actions under human review.
Read more →
Apple’s C2 modem reaches nearly every iPhone 18 Pro, with the U.S. Pro Max as the only documented regional exception.
Read more →
Anthropic’s proposed AI slowdown starts with permanent outside evaluators and expands to domestic safety checkpoints and global coordination.
Read more →
OpenAI's Habitat storage disclosure offers practical lessons in tail latency, connection pooling, constrained APIs, centralization, and staged rewrites.
Read more →
California's new laws establish qualification, registration, independence, and reporting rules for AI auditors without imposing a universal audit mandate.
Read more →
iPhone Duo brings Apple’s first foldable screen, while iPhone 18 Pro offers stronger camera tools for $800 less. Compare price, battery, release dates, and the practical tradeoffs.
Read more →
Series 12 and Ultra 4 share Apple’s new health sensing and readiness features, but Ultra 4 doubles the starting price in exchange for much longer battery and outdoor tracking.
Read more →
Anthropic’s new alignment assessment adds a fourth real-system incident and reframes the failures as both containment errors and model misalignment.
Read more →