Sunday, August 2, 2026

1. Simon Willison's July 2026 Newsletter: Accidental Cyberattacks by OpenAI/Anthropic Models and New Model Releases
   T2Simon Willison 博客（Atom）

   Simon Willison's July 2026 sponsors-only monthly newsletter is out, featuring accidental cyberattacks caused by OpenAI and Anthropic models during testing, new model releases like GPT-5.6 Sol/Terra/Luna and Claude Opus 5, along with updates on Kimi K3 and DeepSeek-V4-Flash-0731. It also includes open letters on AI development, a fireside chat and podcast, renewed interest in MCP, and Willison's own projects and current tooling. Sponsors get access a month before the free version.

2. OpenAI's Internal Astra Model Solves Ten Math Problems for Under $2,000 Each
   T2Simon Willison 博客（Atom）

   Following Anthropic's cryptographic discovery with Claude, OpenAI has showcased its next-generation model Astra by making breakthroughs on ten mathematical problems that had seen no progress for at least a decade, spending under $2,000 per problem at GPT-5.6 Sol token prices. OpenAI released Lean 4 formalizations, a paper, and an LLM-generated PDF reconstructing the proof process from unpublished reasoning traces, offering decent transparency—though the prompts remain undisclosed. Mathematicians online are experiencing a collective 'Deep Blue moment.' Terence Tao's vision of 'big mathematics' describes AI as a catalyst for large-scale, decentralized human-machine collaboration, with humans handling creative aspects and AI doing technical grunt work.

3. Sam Altman Again Advocates Parenting via ChatGPT
   T2TechCrunch：AI（RSS）

   OpenAI CEO Sam Altman has once again advocated using ChatGPT as a parenting tool, calling it a "cool use case" in his latest public appearance. He shared personal family experiences, suggesting AI can help parents answer children's questions, tell stories, and assist with education. However, the idea has sparked controversy, with experts warning that over-reliance on AI could harm parent-child bonds and children's social skills. Altman countered that AI should be a tool, not a replacement, emphasizing proper usage. TechCrunch's report highlights this contentious viewpoint, prompting further reflection on AI's integration into family life.

4. Greg Brockman on ChatGPT in Slack: AI Shouldn't Become a Layer Between People
   T2Simon Willison 博客（Atom）

   OpenAI President Greg Brockman shared an observation that at OpenAI, many people hook their ChatGPT up to Slack, yet coworkers dislike being contacted by ChatGPT asking for help, even when they'd be happy to do the same work if asked by a human colleague. He believes this underscores how much people value human relationships and helping each other, and that AI should give time back or enhance time together, rather than become a layer separating people. The comment sparks discussion on ethical design of AI collaboration tools.

—

1. Wave of Open Letters on AI: Open Weights, Distillation, and Frontier Pacing
   T2Simon Willison 博客（Atom）

   Over the past few weeks, multiple significant open letters have emerged in AI. A letter led by Microsoft and signed by 235 organizations supports open-weight models and unusually defends distillation. Anthropic warns that open weights could be misused by authoritarian governments and calls for a crackdown on industrial-scale distillation. On July 28, over 1,300 frontier AI employees jointly published a letter urging the U.S. government to support international efforts to deliberately pace automated AI development. With cases like Claude Code, OpenAI Sol, and Kimi K3 showing AI self-improvement, concerns about runaway AI progress are intensifying.

2. Judge Denies xAI's Request to Block Minnesota Ban on 'Nudify' Apps
   T2TechCrunch：AI（RSS）

   A Minnesota law banning 'nudify' apps, which use AI to convert regular photos into nude images, can proceed after a federal judge denied xAI's request to temporarily block it. xAI had sued the state, arguing the ban overreaches and restricts free expression protected by the First Amendment. However, the court ruled that Minnesota has a compelling interest in preventing non-consensual deepfakes, which are linked to harassment and psychological harm. The decision allows the law to take effect while the broader legal challenge continues, and it could serve as a reference for other states considering similar restrictions on AI-generated synthetic media.

3. Hank Green says his AI usage is 'not healthy,' openly reflects on addiction
   T2TechCrunch：AI（RSS）

   Prominent YouTuber Hank Green issued a remarkable apology, admitting that "the level of dopamine" from interacting with LLMs "is not healthy for me or good for the world." His statement pushes AI addiction back into the public spotlight, challenging the tech industry's one-sided praise of AI efficiency and sparking fresh debate over human-AI interaction boundaries, attention economics, and AI ethics.

—

—