The Information Machine
Today

Saturday 29 August 2026

20Moved
37New this week
146On the record

Top stories

01
Day 9

Anthropic's revenue surge and IPO push

  • Anthropic's IPO window narrowed on August 28 to late September or early October, with a prospectus expected after Labor Day; the company plans to pitch on a total addressable market of more than $30 trillion and is considering lockups longer than 180 days for some shareholders.
  • Reporting on August 29 added that Meta was projecting up to $10 billion annually on Anthropic's tools, even as Mark Zuckerberg published a 6,500-word piece taking implicit shots at rival AI labs.
The gist

An IPO at a near-$2 trillion valuation would rank among the largest on record. The revenue trajectory and Meta's projected spend provide concrete signals of Anthropic's commercial scale heading into the offering.

Continue this story →13 sources
02
Day 4

Meta's child addiction litigation

  • Meta agreed on August 28 to pay between $12.1 billion and $17.1 billion to settle a 47-state lawsuit claiming it deliberately designed its platforms to be addictive to children, with terms including a two-hour daily limit, age assurance subject to false-positive-rate caps, and teens given the option to disable personalized feeds.
  • A Los Angeles jury had already returned a $6 million verdict in March 2026, and four states opened an Oakland trial in mid-August 2026 seeking up to $1.4 trillion in penalties.
The gist

The 47-state settlement imposes specific, independently tested behavioral restrictions on how Meta serves minors, with concrete error-rate caps on age verification. A separate trial seeking penalties near Meta's full market capitalization means the company's legal exposure on minor-safety claims is not fully resolved.

Continue this story →4 sources
03
Day 4

AI tools compressing software exploit timelines

  • Palo Alto Networks' Unit 42 is investigating an agentic attack that exploited 50 vulnerabilities within one organization in 10 hours, consistent with a METR analysis published August 29 finding that 2026 vulnerability rates for cURL, OpenSSL, Firefox, and Microsoft dramatically outpaced 2025, a shift the authors frame as a 'phase change' in AI cybersecurity capabilities.
  • Automated agents probed an OCaml project within ten minutes of a patch being shared, and rclone received over 40 disclosures in a month against 20 across its first decade.
The gist

The acceleration affects multiple attack surfaces simultaneously: automated probing of public repositories now begins within minutes of patch disclosure, disabling traditional embargo windows; state-level adversaries are integrating AI tools to scale attack volume; and CVE assignment infrastructure is already backlogged. Open-source security practices built around days-long response windows no longer match the actual threat timeline.

Continue this story →4 sources
04
Day 7

Z.ai GLM-5.3 versus frontier coding models

  • Three independent comparisons published August 28 put Z.ai's GLM-5.3-Flash at roughly 17x less cost than the full GLM-5.3, with the pass@1 gap narrowing from 5.6 to 2.6 points at pass@4 in a 900-rollout DeepSWE test and 16.7x less on a Blender MCP task.
  • A three-model price test placed Flash at $0.03 versus $0.29 for Claude Opus 4.6 on the same prompt.
  • Rohan Paul found that wrapping full GLM-5.3 in the Atomic Agent harness roughly doubled token usage while adding only $0.77 to cost.
The gist

Multiple independent benchmarks show a 320B open-weight model running on Chinese chips matching or beating closed frontier models on coding at a fraction of the cost, with detailed cost-quality tradeoff data now available across both the full model and its Flash variant. Bloomberg Intelligence has flagged Z.ai's unit economics as unsustainable given rising inference costs.

Continue this story →13 sources
05
Day 7

AI authorship disclosure at major newspapers

  • Semafor on August 28 named the New York Times, Washington Post, and Wall Street Journal as outlets that regularly publish AI-generated guest columns, having analyzed every guest column the three ran over the past month; both the NYT and WaPo explicitly prohibit AI-generated writing.
  • Semafor concluded blanket bans will become unenforceable as AI detection grows less reliable.
  • Pew's finding that one-third of post-ChatGPT webpages show AI authorship signs came from analyzing nearly 500,000 pages using a tool called Open Pangram.
The gist

Major U.S. newspapers are publishing AI-generated guest columns in apparent violation of their own stated policies, raising direct questions about editorial enforcement and disclosure norms. Pew Research data show AI-associated writing patterns have spread substantially across the commercial web since ChatGPT's launch.

Continue this story →4 sources

Also moved today

Still open, no news today

The daily email

Want this in your inbox?

I send a short email each morning with the stories that moved. If you would rather just read here, that works too.

Subscribe free