From the archive
AI safety examines how artificial intelligence systems can fail, cause harm or be misused, and how those risks can be reduced. Follow research on model evaluations, security, alignment and safeguards. Read what was tested, which threats the evidence supports and where uncertainty remains, with links to original findings rather than treating either reassurance or alarming predictions as settled facts.
Lead story
The Wall Street Journal’s account traces the long-running personal and philosophical conflict between Anthropic chief executive Dario Amodei and OpenAI chief executive Sam Altman to disagreements over AI safety, leadership and credit. It matters because those arguments now sit behind two companies whose decisions help shape the AI…
Read the report & discussion ↗Watch Desk
Scottish mid-sized businesses say AI is helping them serve customers faster, but many are not seeing those efficiency gains translate into better finances. A survey for FRP Advisory suggests that rising costs and growing customer expectations are absorbing much of the benefit.
Watch Desk
Some influencers are promoting cheaper camera- and AI-equipped glasses as a way to record people without an obvious warning light, Futurism reports. The pitch turns a privacy safeguard into a selling point, though the article says some claims about the glasses’ lights are disputed.
Yoshua Bengio Watch
Yoshua Bengio is urging researchers who put AI safety first to leave frontier AI companies and join organisations working on safer systems. In a personal essay published on 8 October, the AI researcher argues that competition between leading labs is working against their safety efforts, and says researchers…
Watch Desk
Abu Dhabi has launched AiNative, a programme aiming to train 20,000 government employees in AI by the end of 2026. The practical point is that the training is being tailored to people’s jobs, with guidance on what government information may be entered into AI systems.
Watch Desk
Ten major AI developers have changed, or promised to change, how they handle personal data after scrutiny by the UK’s Information Commissioner’s Office (ICO). The regulator is also examining whether autonomous AI agents can follow the rules when they act with less human supervision.
Watch Desk
Mark Zuckerberg helped shape a voluntary AI safety accord signed by technology executives at the White House, CNN reports. The episode puts a practical question at the centre of AI policy: who gets to write the rules when the rules are voluntary?
Watch Desk
AI trading startup Catalyst has raised a $30 million seed round led by Sequoia, Fortune reports. The company says a pilot generated hundreds of millions of dollars in trading volume over two weeks, a striking figure that measures activity, not whether investors made money.
Watch Desk
Jeff Bezos says AI-driven productivity could let some people support their families on a single income or work a three-day week. In a Fox News interview, he also argued for light regulation of AI while the technology develops, putting a prominent tech leader’s optimistic view of work and…
Watch Desk
SafeWorld has emerged from stealth with $12.2 million to build software that tests how autonomous robots behave in dangerous or unexpected situations. The company says its simulations can expose more edge cases before robots are deployed around people, though the announcement offers no independent results showing how well…
Watch Desk
A malware campaign called PoeLLM is targeting exposed AI and development servers, using compromised machines to spread and mine cryptocurrency, according to BleepingComputer. The report says the activity has been running since at least April, putting a practical security question in front of anyone running these tools: are…
Watch Desk
Singapore’s financial regulator wants firms to have every AI use case independently reviewed before it goes live. The guidance also calls for ongoing monitoring, including of third-party AI, and says firms remain accountable for AI used in the services they provide.
OpenAI Watch
OpenAI says it has launched GPT-6 Sol and GPT-6 Luna globally for free and paid ChatGPT users, replacing GPT-5.6 models. Its safety hub classifies both as high capability in cybersecurity and biological and chemical domains, a consequential designation alongside the company’s claimed safety improvements.
Watch Desk
As AI coding agents speed up software development, engineering leaders are rethinking team structures, safety practices and what skills developers need. At QCon London 2026, Hannah Foxwell argued that when code gets cheaper to produce, choosing worthwhile work and keeping people and systems safe become harder parts of…
Watch Desk
Former White House AI adviser Sriram Krishnan is targeting a $500 million venture fund for growth- and later-stage AI companies, The Information reports. If raised, it would put a substantial pool of capital behind companies further along the road from promising pitch to established business.
AWS AI Watch
AWS says Amazon Quick and Amazon Bedrock Knowledge Bases can now check document permissions against their authoritative source at query time, rather than relying only on access data copied during an earlier sync. The practical gain is that revoked access can be reflected in AI answers within moments,…
Microsoft AI Watch
Microsoft says its FORGE Lab used AI-assisted research to find 140 Windows vulnerabilities between May and September 2026, including 52 addressed in September’s security release. The company’s account also describes a growing open-source vulnerability programme, where the work of validating and fixing findings may be harder to scale…
Watch Desk
AI leaders who fear the consequences of powerful systems may also feel compelled to build them before less cautious rivals do, journalist Kevin Roose argues. His proposed counterweight is a mix of oversight and coordination, not simply hoping the race slows itself down.
