A team at OpenAI has devised an approach that gets their AI systems to produce confessions—self-generated explanations where the model reflects on its actions and admits to any questionable conduct. Understanding misleading behaviours in large language models – such as hallucination, dishonesty or manipulation – has become one of the most
The European Commission has opened an antitrust investigation into Meta over its WhatsApp AI policy, examining whether the company's restrictions on third-party AI providers breach EU competition rules. Meta announced in October a policy restricting AI providers's access to contact customers via WhatsApp. The investigation will
French AI startup Mistral launched its new Mistral 3 family of open-weight models on Tuesday, a release that presents a different focus than that of its big Western tech rivals. The 10-model release includes a large frontier model with multimodal and multilingual capabilities and nine smaller offline-capable, fully customisable models.
Perplexity has released BrowseSafe, an open research benchmark and content detection model designed to protect users as AI agents integrate into web browsers. Prompt injection has emerged as a critical attack vector for web agents, yet its real-world impact remains insufficiently understood. This release addresses a significant gap in AI
Artificial intelligence could widen inequality between countries by increasing divides in economic performance, capabilities, and governance systems, according to a United Nations Development Programme report titled The Next Great Divergence: Why AI May Widen Inequality Between Countries, released on 2 December 2025. Whilst AI opens important avenues for development, countries
China's DeepSeek and Alibaba-backed Moonshot have released new AI models that demonstrate competitive performance against leading Western proprietary systems. DeepSeek launched V3.2 in September and V3.2-Speciale in December, whilst Moonshot introduced Kimi K2 in July and Kimi K2 Thinking in November. Both companies claim that their
With holiday shopping on the horizon, OpenAI and Perplexity both announced AI shopping features integrated into their existing chatbots to help users research potential purchases. ChatGPT's shopping research began rolling out on mobile and web for logged-in users, whilst Perplexity's feature launched free for all US
In mid-September 2025, Anthropic detected suspicious activity that investigation determined to be a sophisticated espionage campaign in which a Chinese state-sponsored group manipulated Claude Code into attempting infiltration of roughly 30 global targets, including large tech companies, financial institutions, chemical manufacturing companies, and government agencies. Anthropic believes this is the
On 24 November 2025, President Donald J. Trump signed an Executive Order launching the Genesis Mission, a national effort to use AI to transform scientific research and accelerate discovery, comparing the initiative in urgency and ambition to the Manhattan Project. The Department of Energy will lead the effort by leveraging
Yann LeCun, one of the godfathers of modern AI, is departing Meta at the end of the year after 12 years with the company, where he served as founding director of Facebook AI Research (FAIR) and Chief AI Scientist. LeCun plans to launch a startup focused on world models, a
Warner Music Group announced that it has reached settlements with AI music startups Suno and Udio, resolving copyright infringement lawsuits filed just over a year ago by the Recording Industry Association of America (RIAA) on behalf of major record companies. The deals mark a significant shift from litigation to collaboration,
On 9 November 2025, Mixpanel, a third-party web analytics provider used by OpenAI for its API product frontend, suffered a security breach when an attacker gained unauthorised access to part of their systems and exported a dataset containing limited customer identifiable information and analytics data. OpenAI confirmed the incident affected
