Twelve of the 247 security fixes in Chrome 155 carry the line "assisted by Claude", two of them critical

The thirtieth of September. Xinyang Ge of Anthropic reports a critical bug in Chrome, a use after free in the Track component, and in the bulletin the name is followed by "assisted by Claude". On 6 October the fix is in stable Chrome 155. Six days.

Twelve lines with the same signature in a bulletin of 247 fixes. All twelve are use after free, a bug where a program uses memory it has already released. Of the four critical fixes in this version, two have Claude next to the researcher's name, and one belongs to Google itself.

How much of the work is the human's and how much the model's, the bulletin does not say. It says "assisted by", meaning with the help of. Two more lines carry the name Xinyang Ge without Anthropic and Claude next to it, among them a third critical fix. We do not count those toward Claude, because Google does not mark them that way either.

Twelve lines in Google's bulletin, and every one says "assisted by Claude".

The same list also has two fixes signed OpenAI Codex Security, in V8 and in HTML. Anthropic and OpenAI in one Google bulletin. The rest is more familiar: 185 of the fixes, close to three in every four, were reported by Google itself.

There is no urgency this time. For none of the 247 does Google say it is being used in attacks, and Chrome updates itself, as long as you restart it.

Whether it is a direction, one number will tell: how many lines with "assisted by" the next big Chrome bulletin carries, and they are easy to count.

The visual is generated code-art, no third-party images.