- 01
The agency added both PHP flaws to its Known Exploited Vulnerabilities catalogue on 26 October 2026, a week after saying it had seen no successful exploitation, and gave federal civilian agencies until 29 October to act.
Why it matters PHP runs the server-side logic of most of the web, so a catalogued unauthenticated code execution flaw in it converts into a patching obligation across government and a scanning target for everyone else inside the same week.
3 min read - 02
UN Climate Change synthesised 127 new national pledges covering 71 per cent of global emissions and found they would cut output 6 per cent below 2019 levels by 2035. The same morning the WMO put atmospheric carbon dioxide at a record 426.8 parts per million.
Why it matters COP31 opens in Antalya on 9 November with a pledged 2035 cut of 6 per cent against a scientific requirement near 60 per cent. Every finance, adaptation and fossil fuel argument at that conference is downstream of this arithmetic, which is now on the official record rather than in advocacy estimates.
3 min read - 03
GPT-6.1 Terra lists at $1 per million input tokens, half the rate of the model it replaces and of the stronger GPT-6.1 Sol. Artificial Analysis scored it 48 on its Intelligence Index, four points behind Sol.
Why it matters Agent economics have been gated by per-token cost rather than capability. A mid-tier model that holds its measured score while cutting cost per task from $1.40 to $0.40 moves the threshold at which routine screen-based office work becomes cheaper to automate than to staff.
3 min read - 04
OpenAI published three claimed results on 26 October 2026, including a refutation of a conjecture that hundreds of complexity papers assume to be true. None has been peer-reviewed, and the group checking the refutation is led by the two men who proposed the conjecture.
Why it matters If the refutation of the Exponential Time Hypothesis survives scrutiny, hundreds of published hardness results that assumed it become conditional on nothing. The procedural shift matters as much: OpenAI is releasing machine-generated proofs as raw data and asking unpaid academics to work out what they mean, which puts the verification burden on a community with no agreed method for doing it.
4 min read - 05
The Claude developer applied on 26 October 2026 to list on Nasdaq as ANT. Revenue for the six months to 30 June was 14 times the year before, and almost 90 per cent of the $80.7bn net loss is a non-cash charge for the rising value of its own convertible notes.
Why it matters Anthropic's prospectus is the first detailed public look inside a frontier AI developer's economics, and it shows a business that turned an operating profit in one quarter while signing compute commitments that run to 2036. The mismatch between fixed obligations and cancellable revenue is the clearest public measure yet of how much of the AI build-out rests on contracts rather than on demand.
4 min read