MartinMacdonald
Member
- Joined
- May 13, 2026
- Messages
- 1
- Thread Author
- #1
The Claude Mythos Preview findings change the ESU argument completely. When Microsoft set October 2026 as the Windows 10 ESU end date, no one knew that an AI model would shortly identify thousands of zero-day vulnerabilities across every major OS and browser — including decades-old bugs that survived years of human review.
Anthropic's CEO has publicly said there's a 6–12 months window to patch these before adversarial AI catches up.
For consumers on hardware that can't run Windows 11 and hardly can afford to purchase new computers, ESU ending in October 2026 doesn't mean a "gradually rising risk" anymore. It means permanent and high probability exposure to vulnerabilities bad actors using AI can find and exploit autonomously, in minutes. Schools, small businesses, health clinics, households on fixed incomes (that’s my wife and me), these are exactly the targets researchers say will be hit first and hardest.
Project Glasswing launched by Anthropic helps companies identify and patch software vulnerabilities by providing early access to its advanced AI model, Claude Mythos Preview.
This initiative aims to secure critical software by enabling defenders to find security flaws faster than attackers can exploit them.
Microsoft could move the ESU end date to ensure at least critical risks are mitigated, if not all risks. If nothing else, litigation risks should influence their decision.
European consumer groups already moved Microsoft once on ESU terms. The Mythos findings are a far stronger argument.
Anyone else think it's time to push again?
Anthropic's CEO has publicly said there's a 6–12 months window to patch these before adversarial AI catches up.
For consumers on hardware that can't run Windows 11 and hardly can afford to purchase new computers, ESU ending in October 2026 doesn't mean a "gradually rising risk" anymore. It means permanent and high probability exposure to vulnerabilities bad actors using AI can find and exploit autonomously, in minutes. Schools, small businesses, health clinics, households on fixed incomes (that’s my wife and me), these are exactly the targets researchers say will be hit first and hardest.
Project Glasswing launched by Anthropic helps companies identify and patch software vulnerabilities by providing early access to its advanced AI model, Claude Mythos Preview.
This initiative aims to secure critical software by enabling defenders to find security flaws faster than attackers can exploit them.
Microsoft could move the ESU end date to ensure at least critical risks are mitigated, if not all risks. If nothing else, litigation risks should influence their decision.
European consumer groups already moved Microsoft once on ESU terms. The Mythos findings are a far stronger argument.
Anyone else think it's time to push again?