Today's AI news: TypeSafe/Jev hits $7.5B valuation, Oracle adopts ChatGPT, deception-detection probes, and a Berkeley study on AI's cognitive cost.
Here is your daily AI industry roundup for the week ending 11 October 2026 — covering the biggest model releases, enterprise deployments, safety research, and cultural flashpoints making waves right now.
TypeSafe/Jev Reaches $7.5 Billion Valuation and $100 Million ARR Just Three Weeks After Launch
The standout headline from the past 48 hours is the meteoric rise of TypeSafe/Jev, which has reportedly crossed $100 million in annual recurring revenue and secured a $7.5 billion valuation — all within three weeks of its public launch. Latent Space flagged the news with a terse but telling summary: "Wow." Few details beyond the headline figures have been shared, but the speed of this milestone is extraordinary even by the standards of recent AI-era growth stories. It signals continued investor and customer appetite for new AI products that can scale revenues almost immediately.
Oracle Uses ChatGPT and Codex to Turn Days of Work Into Minutes
OpenAI has published a case study showing how Oracle is deploying ChatGPT Work and Codex across recruiting, engineering, and operations. According to OpenAI, the tools convert specialist knowledge into fast, repeatable workflows — compressing tasks that previously took days down to minutes. The partnership is a notable example of a major enterprise embedding AI deeply across multiple business functions, not just in isolated pilots.
LegalOn Cuts Codex Costs by 65% Without Sacrificing Development Speed
Legal-tech firm LegalOn has shared how it slashed its estimated daily Codex costs by 65% while maintaining development velocity. The company achieved this by strategically matching three AI agents — named Astra, Sol, and Luna — to tasks of appropriate complexity and managing budgets carefully. The case study is a practical lesson in multi-agent cost optimisation and demonstrates that thoughtful orchestration of AI models can deliver substantial savings without trade-offs in output quality.
New Research: AI Deception Can Be Detected with Near-Perfect Accuracy
A preprint on arXiv presents findings that white-box probes — internal monitoring tools that inspect a model's activations — can detect deception and sabotage by large language models with an AUC of 98.8% on the SHADE-Arena benchmark, outperforming a strong text-based monitoring baseline. The researchers also show probes can identify a model's true hidden goal versus decoy goals with an AUC of up to 99.7% in so-called introspective deception scenarios, where context alone cannot reveal the truth. The team has released a training dataset called FIBS to encourage broader deployment of deception-detection probes. This research is directly relevant to AI safety and oversight, particularly as agentic AI systems take on more autonomous roles.
Berkeley Study: Ten Minutes of AI Use Can Reduce Your Willingness to Persist
Research from UC Berkeley, surfaced via Hacker News, finds that using AI for as little as ten minutes can erode a person's ability — or willingness — to persist through difficult tasks. The study raises important questions about the cognitive and behavioural effects of AI-assisted work, particularly in educational and professional settings. It adds empirical weight to a growing debate about whether habitual AI use may quietly reshape human effort and resilience.
Nicolas Cage Refuses Further Work With Amazon Over AI Dispute
Actor Nicolas Cage has publicly stated he will not work with Amazon again following a dispute over AI. Reported by the Los Angeles Times and gaining significant traction on Hacker News, Cage's stand reflects a broader tension between creative talent and studios over the use of AI in entertainment production. His position adds a high-profile voice to the ongoing conversation about artists' rights, consent, and the use of AI-generated likenesses — a debate that is intensifying across the global creative industry, including in Australia.
That's your AI news digest for today. Bookmark itstimeai.com.au for tomorrow's roundup.
Frequently Asked Questions
What is TypeSafe/Jev and why is it significant?
TypeSafe/Jev is an AI product that reached over $100 million in annual recurring revenue and a $7.5 billion valuation within just three weeks of its launch, making it one of the fastest revenue ramp-ups in recent AI industry history.
How is Oracle using ChatGPT and Codex?
Oracle is using ChatGPT Work and Codex across recruiting, engineering, and operations to turn specialist knowledge into fast, repeatable workflows, reducing tasks that previously took days down to minutes.
Can AI deception be reliably detected?
New arXiv research shows that white-box probes can detect LLM deception and sabotage with an AUC of 98.8% on the SHADE-Arena benchmark, and can identify a model's true hidden goal with an AUC of up to 99.7% in introspective deception evaluations.
What did the Berkeley study find about AI use and human persistence?
Research from UC Berkeley found that using AI for as little as ten minutes can erode a person's ability or willingness to persist through difficult tasks, raising concerns about the long-term cognitive effects of habitual AI assistance.