The Future of AI Research report, published in March 2025, aims to clearly identify the trajectory of AI research in a structured way. The report was led by outgoing AAAI President Francesca Rossi and covers 17 different AI topics . Members of the report team, and other selected AI practitioners, are taking part in a series of video panel discussions covering selected chapters from the report. In the sixth discussion in the collection, the three panellists tackle factuality and trustworthiness. Understanding factuality: why preventing false outputs from large language models remains AI's toughest problem Lucy Smith is Senior Managing Editor for AIhub.

UK fears a'triple whammy': oversized investment in AI stocks, slower adoption of AI than predicted and the breakneck pace of AI's development The UK wants a piece of the mammoth global investment in AI but fears it as well. In the coming weeks, the Bank of England is planning to ease capital rules to help encourage more lending. But the central bank simultaneously expressed concerns that there are too many loans going to investors like hedge funds, who are using that money to buy up AI stocks. The central bank's moves reflect the country's global position: hoping to catch up to the US and China in the AI race, struggling to mobilize its resources to do so, and too wary of the risks to go full bore. UK banking regulators have recently been under enormous pressure to do more to stimulate growth, my colleague Kalyeena Makortoff this week reports.

Plus: New York has become the first state to enact a data center moratorium. When Anthropic announced last week that it had found a new window into its models' "internal thoughts" as they reason through answers, there was one colleague I had to talk to: senior editor Will Douglas Heaven. Aside from having a PhD in computer science, Will has spent a lot of time digging into what we can say about how AI models work. I spoke with him about what we should take from Anthropic's new (and typically quirky) research. Here's what he had to say . How will AI understand the real world?

Check your subscription status, update your details and more. Powerful artificial intelligence models built by Chinese companies have gone from inducing widespread panic to being met with a shrug of the shoulders - what changed? When Chinese company DeepSeek released its open-source R1 model in January 2025, it made headlines around the world. The large language model (LLM) was reported to rival some of the most powerful AIs from US companies, but it was completely free for anyone to download. A trillion dollars was wiped off the value of US tech companies and US lawmakers immediately proposed banning it on government devices.

You'll start seeing the'Ask Gemini' button in your browser. Google has just rolled out Gemini for Chrome in the UK. The integration doesn't sit well with everybody -- and Gemini isn't the most popular AI assistant out there -- but you will start seeing an Ask Gemini button with a sparkle icon soon if you're in the United Kingdom and use Google's browser. Gemini's Chrome integration used to be exclusively available to AI Pro or AI Ultra subscribers, but Google made it more widely available in the US on desktop in late 2025. The company then rolled out the feature to Latin America and the Middle East, and then to Canada, India and New Zealand, until it made its way to over 50 countries around the world.

Who will win the World Cup? Who will win the 2026 FIFA World Cup? Here's what AI predicts As the 2026 FIFA Men's World Cup enters its final stages, AJLabs asked nine leading AI models to predict the tournament's final podium based on all available data for each team, including: France emerged as the favourite to lift the trophy, receiving five (Gemini, Grock, DeepSeek, Le Chat and Qwen) of the nine champion votes. Argentina, the defending world champions, received the remaining four votes (ChatGPT, Claude, Copilot and Meta AI). Predictions for the runner-up were more divided: France and Argentina each received three votes, followed by England with two and Spain with one. Spain was the clear favourite to finish third, receiving six of the nine third-place predictions, while England and France each received fewer votes. The predictions reflect a broad AI consensus around the four remaining contenders, France, Argentina, Spain and England, but also highlight differences in how leading language models weigh recent performances, squad depth and tournament momentum.

Labor MP says'a fair day's pay for a fair day's work' was a founding principle of the ALP as media union calls for tougher new rules on AI use of creative work Husic also urged his colleagues to place stricter rules on the big tech firms or be "doomed to failure". Husic, who has long advocated for a more interventionist approach on AI policy, said big firms like OpenAI and Anthropic should not be left to self-regulate, and that the federal government should be setting strong rules. "If we were to wait for social licence with industry, we wouldn't get emissions reduction. Governments sometimes have to step in," Husic told Sky News on Tuesday. Going down the path of social licence with tech is a path that's sadly doomed to failure, because we tried self-regulation for a couple of decades and found out that it didn't work." The prime minister will deliver a highly anticipated speech in Sydney on Wednesday to address growing concerns around social licence and the necessary policy guardrails for AI, datacentres and Australian intellectual property. We've grown up with the notion of a fair day's pay for a fair day's work - that people should be remunerated fairly for the labour, the effort that they provide. Asked whether he thought his colleagues were doing that, Husic replied: "Obviously, there's a debate that's going on behind the scenes.

Inside Story How is the US tech industry regulated? How is the US tech industry regulated? Apple is taking on OpenAI in court. The tech giant has filed a lawsuit against the artificial intelligence company for what it calls pervasive theft. OpenAI has denied the accusations.

The company says it has found a new window into how its models arrive at answers. We spoke with senior editor Will Douglas Heaven about it. Anthropic--currently the world's most valuable AI company, with a nearly $1 trillion valuation--has a reputation for publishing strange and heady research. It's looking into whether AI models can feel pain, for example, and will sometimes cut off chatbot conversations if it suspects users are "abusing" the model. One niche that Anthropic spends more time and money on than other AI companies is called mechanistic interpretability, which means looking inside the complex math of an AI model to learn why it comes up with one particular output and not another. It's complicated stuff; there are millions of data points that might contribute to any result, and wading through them can look more like word salad than anything useful.

PCWorld reports that OpenAI and Anthropic have temporarily relaxed usage limits for their latest AI models, including GPT-5.6 Sol and Claude's Fable 5. ChatGPT Plus, Business, and Pro subscribers can now access GPT-5.6 Sol without the previous five-hour usage window restrictions. This increased access appears driven by competition between the AI companies, giving users more opportunities to experience advanced reasoning capabilities. If you haven't kicked the tires yet on the latest and greatest ChatGPT and Claude models, this is your lucky week. OpenAI is (fittingly) opening the flood gates to GPT-5.6 Sol, its just-released and most powerful model, announcing Sunday that it's "temporarily" lifting the five-hour usage window for ChatGPT Plus, Business, and Pro subscribers. At the same time, Anthropic is - again - extending the trial period for Fable 5, its own new top-of-the-line, giving Claude subscribers another week of in-plan access .