Google boss says research needed on AI threats, Microsoft confirms private emails read by Copilot AI

Share

More research on the threats of artificial intelligence (AI) “needs to be done urgently”, the boss of Google DeepMind has told BBC News. In an exclusive interview at the AI Impact Summit in Delhi, Sir Demis Hassabis said the industry wanted “smart regulation” for “the real risks” posed by the tech. Many tech leaders and politicians at the Summit have called for more global governance of AI, ahead of an expected joint statement as the event draws to a close. But the US has rejected this stance, with White House technology adviser Michael Kratsios saying: “AI adoption cannot lead to a brighter future if it is subject to bureaucracies and centralised control.” BBC

Elon Musk is suing the EU over a landmark €120m (£105m) fine against his social media company X, accusing Brussels officials of bias. X launched an appeal against December’s fine at the EU General Court earlier this week, in what is the first legal challenge to Europe’s tough digital laws. The challenge will escalate a row between Mr Musk, Brussels and the White House after the Trump administration claimed EU policies are suppressing free speech. Telegraph 

Mind is launching a significant inquiry into artificial intelligence and mental health after a Guardian investigation exposed how Google’s AI Overviews gave people “very dangerous” medical advice. In a year-long commission, the mental health charity, which operates in England and Wales, will examine the risks and safeguards required as AI increasingly influences the lives of millions of people affected by mental health issues worldwide. Guardian 

An ongoing phishing campaign that targets Microsoft 365 users by abusing OAuth tokens to gain long‑term access to corporate data, which focuses on business users in North America and aims to compromise Outlook, Teams, and OneDrive without directly stealing passwords. Instead of attacking login pages with fake forms, the operators trick victims into completing a real sign‑in process on Microsoft’s own device login portal, which makes the attack harder for both users and basic security tools to spot. Cybersecuritynews 

Microsoft has confirmed that a bug allowed its Copilot AI to summarize customers’ confidential emails for weeks without permission. The bug, first reported by Bleeping Computer, allowed Copilot Chat to read and outline the contents of emails since January, even if customers had data loss prevention policies to prevent ingesting their sensitive information into Microsoft’s large language model. Copilot Chat allows paying Microsoft 365 customers to use the AI-powered chat feature in its Office software products, including Word, Excel, and PowerPoint.  Techcrunch 

If you want an even better AI model, there could be reason to celebrate. Google, on Thursday, announced the release of Gemini 3.1 Pro, characterizing the model’s arrival as “a step forward in core reasoning.” Measured by the release cadence of machine learning models, Gemini 3.1 Pro is hard on the heels of recent model debuts from Anthropic and OpenAI. There’s barely enough time to start using new US commercial AI models before a competitive alternative surfaces. And that’s to say nothing about the AI models coming from outside the US, like Qwen3.5. TheRegister 

For latest tech stories go to TechDigest.tv