RE: https://flipboard.com/@404media/404-media-qvt3vv94z/-/a-WUaas_e8S32MEVyncTs4mA%3Aa%3A4082434389-%2F0

This is a great illustration about the state of the art in "AI": People with access to inconceivable resources running massively expensive, dangerous code on anyone's data (sometimes even their own) with no regards for basic security or quality standards and practices.

The discipline of software engineering has never been great at establishing and following those kinds of practices but "AI" has thrown us back decades. And we'll suffer the drop in software quality and security for a long time.

@tante need an account to read it ):
@criticalbackend well, I can recommend subscribing to 404 media (even the free tier) but https://www.businessinsider.com/meta-ai-alignment-director-openclaw-email-deletion-2026-2 also has the screenshots and info
Meta AI alignment director shares her OpenClaw email-deletion nightmare: 'I had to RUN to my Mac mini'

Meta alignment director Summer Yue hooked OpenClaw up to her inbox. Then, the bot tried to delete her emails. Yue chalked it up to a "rookie mistake."

Business Insider
@tante Most software development has little or no "engineering", if you're even remotely disciplined in using that word. Engineering means being held to a standard, and being legally liable for lapses. But the EULA made sure that software companies need never bother with such niceties. This is a fundamental mistake.

@tante @404media

Reminds me of when Microsoft first released internet-capable systems.

They recapitulated every buffer-overflow lesson learned on Unix, as well as a few of their own.