Microsoft’s AI journey has been a challenge and a new bug has raised further privacy concerns for millions. The company faced a situation where the AI summarise feature was somehow able to access private
emails of its users bypassing the security of companies. The details covered by Bleeping Computer in its report once again shows the rush to roll out new AI tools cannot happen at the cost of user privacy and their data.
The bug was found in Copilot AI which was running in the enterprise version and the AI was able to process private emails which it was never supposed to.
AI Reads Your Emails, What Next?
The bug was serious enough for Microsoft to flag, report and act on immediately but even then some companies got the memo a little late and this meant the AI summarise feature ended up pulling up emails from the Sent items and Draft folder, the report adds. The company, quoted by the media house, claims it started rolling out the fix for the bug earlier this month but it has refused to share the impact of the issue on its customers and how many might have seen their data accessed by Copilot AI.
Incidents like these are not helping AI’s cause, especially for Microsoft which is finding it tough to sell its AI Windows 11 laptops and PCs in the market. The company is even looking to tone down the AI push to make people realise the upgrade is more about performance than the gimmicky stuff.
AI attacks, especially involving prompt injections have sparked concerns, purely because of what AI agents can execute, including handling your payments data. Google has faced its own AI glitches, which probably makes Apple feel their delayed AI push isn’t the worst thing to happen.
Copilot, ChatGPT, Gemini and many more AI tools are vying for your data, access to your device and much more. But the AI needs to be controlled, more so in cases where the features are being rolled out without any clearance checks.
