Google has said that AI tools helped it patch 1,072 security vulnerabilities in Chrome across the two versions released in June, exceeding the 1,036 bugs fixed over the previous 23 versions spanning the past two years. Chrome engineering director Doug Turner said large language models have transformed vulnerability discovery into an industrial-scale operation, and that applying models like Gemini allows the company to fix flaws preemptively and stay ahead of attackers. Microsoft reported a similar trend earlier this month, patching a record 570 vulnerabilities across its products and crediting its own AI use, while Apple has not shown the same sharp increase, with independent tracking showing roughly 482 bugs patched so far this year, in line with recent years.
Separately, Google rolled back a newly launched Google Earth feature just one day after release following criticism that it could spread misinformation. The tool let users deploy Google’s Nano Banana 2 image generator to create fabricated, prompt-based imagery superimposed on real satellite maps. Critics, including journalists, warned the feature could undermine Google Earth’s reliability as a source of visual evidence. Google said it had seen geospatial professionals use the tool constructively, but also saw examples of generated imagery that appeared to violate its policies, prompting the company to pull the feature while it develops stronger safeguards.
