On October 29 OpenAI doubled the Work and Codex allowance on ChatGPT Pro 100 and Pro 500, cancelled the planned Pro 200 cut one day before it was due, and raised Business Premium. What each plan gets now, what did not change, why OpenAI says it moved, and what it means for companies and users.
Browsing Tag
OpenAI
24 posts
Who Tests the Most Powerful AI Before It Ships? Britain Pays Far More Than the US, but Not “10 Times” More
Britain just gave its AI Security Institute an extra £120 million, spread over four years. The US testing center at NIST runs on about $15 million a year. Per year, Britain spends roughly 8 to 9 times more, not “10 times” from the top-up alone. What the testers found in 2026, why access matters more than money, how the EU AI Office differs, and what it means for companies and users.
Did China’s Newest AI Models Learn From GPT-6 Astra? What the Distillation Report Shows, and What It Doesn’t
The Information reports that DeepSeek and Moonshot AI relied heavily on GPT-6 Astra outputs to train Kimi K3.1 Pro and DeepSeek V4.1 Flash 1017. One source, no confirmation. What distillation is, what is documented, what is legal, and what it means for companies and users.
Europe’s Governments Are Letting AI In: 25,000 Jobs, Asylum Files and 90,835 Bot Replies
The Netherlands plans 25,000 fewer civil servants with AI as one tool, and warns against overestimating it. Britain uses OpenAI-based tools to summarise asylum interviews without telling applicants. Bots sent 76% of replies to a UK consultation. Sweden blocked public ChatGPT and Claude. What it means for companies and citizens, and your rights.
Attackers Now Go After Your AI Tools: Stolen Sessions, Leaked Agent Logs and Hidden Reasoning
A new infostealer grabs the login files of Claude, ChatGPT Desktop, Codex and Cursor. Warp says a vendor breach exposed agent traces for up to 1% of users. Researchers say OpenAI fixed a flaw that turned hidden reasoning into plain text. What is confirmed, what is not, and a prioritized checklist for companies and users.
OpenAI Says Its AI Proved the Exponential Time Hypothesis False. Here Is What Is Actually Proven
For 25 years, computer scientists bet that a famous puzzle can never be solved much faster than by trying everything. OpenAI says its AI proved that bet wrong and that a machine checked the proof. Explained simply, with everyday examples: what is proven, what is only claimed, why it is not P vs NP, and what it means for companies and users. Full details for the curious below.
Why AI Text Is So Tiring to Read: Six Patterns, and How to Fix Them
A new study counts six patterns that make AI text hard to read: rare word pairs, long gaps, a dropped “that”, noun stacks, nouns as verbs and objects that act alive. Each one with a before-and-after fix, why models drift toward density, why detectors are the wrong tool, and what it means for companies and users. One preprint, honest limits.
Amazon Blocks AI Shopping Agents, Walmart Lets Them In. Who Owns the Customer When an AI Buys?
Meta’s Muse agent shops for 10 million people, Walmart is building a front door for AI agents with spending limits, and Amazon blocks agents it hasn’t agreed to. How agents actually buy, why the real fight is over the customer relationship, what it means for retailers, brands and affiliate sites, and how users keep control.
GPT-6.1 Terra Is Cheap per Task. In Our Newsroom Test, Its Max Setting Was the Worst
OpenAI’s GPT-6.1 Terra costs $1 / $5 per million tokens and sits between Luna and Sol. Artificial Analysis scores it at 48 for $0.40 per task at max effort, against $0.72 for Sol and $5.98 for Claude Opus 5.5, and its six effort settings span a 7-fold cost range. In my own newsroom test (12 articles per language), Terra without reasoning matched GPT-6.1 Sol, and max was its worst setting. Why price per token is the wrong number, and what companies and users should do.
50 Signs of Superintelligence, Checked: How Close Is AI to AGI and ASI?
One researcher counts 50 signs of superintelligence and calls today’s top models early ASI. I traced 18 of his biggest 2026 claims to the primary sources: 7 hold up, 5 are vendor claims, 3 rest on one source, 3 are disputed. Maths is the real breakthrough; AI that improves AI is running, but not exploding. What it means for companies and users.