Drop Site is reporting an integrity breach in its purest form, related to the Intercept leaving a dormant Signal username on a “Become a Source” page. The username was taken over and then a social media account describing itself as “Investigative Intake” solicited tips in roughly 100 posts between February and May, pulling in prospective sources.

Drop Site suggests the takeover exploited Signal’s recycling of dormant usernames. The Intercept framed its username switch as “security best practices”. Best practice actually would be an integrity breach notification, but its chief legal officer declined to address how the account had been seized, how long it persisted, or whether prospective sources were warned.

The Intercept executed a silent rebrand instead of a proper breach notification, while promoting itself to journalists for live training on responding “before, during, and after an online attack“.

Emptywheel sums up the Trump “fair” as built to be unfair.

When you’re guaranteed no-bid contracts to make private profits out of Federal money, you don’t waste time making things look nice. You don’t even waste effort bringing in a full set of carnival rides or live farm animals, the kinds of things that make local fairs so fun.

The fair itself is just the facade, the excuse used to launder millions into private profit.

A social media post by the Lyon County Sheriff’s Office in Nevada makes an almost immediate claim that Tesla Semi design failure is responsible for death.

…statements obtained at the scene suggest the driver of the semi may have fallen asleep.

The 10-ton Tesla robot ignored a red light and hit two cars, destroying one and killing the two people inside. AEB has been mandatory in the EU for a decade for large trucks, whereas in America it’s still optional. It’s not clear yet whether it failed or was missing on this Tesla, which two weeks before the crash was promoting autonomous capabilities.

A thinktank study (Code For Good Now, whose founder is a former Meta strategist) is being pumped by Fortune and Forbes with no published methodology or peer review. But that doesn’t stop it from making some very bold claims.

Women Who Use AI Seen As Incompetent; Men Who Use AI Seen As Pragmatic.

Why are women 25% less likely to use artificial intelligence tools than men? New research debunks the notion that the gender gap is primarily due to women’s lack of AI skills, interest or access. Women’s hesitancy to use AI is instead a rational response to a competency penalty that women face when using AI in the workplace.

Ok, I’ll bite. The Forbes piece helpfully summarizes both a Chatoo study and the 2025 Peking University study (Gai et al.) that it claims to bolster. Putting the two summaries side by side, here is what we have:

1. Gai et al. = bias concentrated in men who do NOT use AI.
2. Chatoo = bias concentrated in men who DO use AI.

Uh, what? So first I’m supposed to read that the penalty dies as adoption spreads, and then I’m supposed to read that adoption is what spreads the penalty?

Perhaps this is why there’s no published methodology or peer review? I was willing to set aside the Meta strategist origin story, but come on.

The “result” of the study is whether the problem self-corrects or requires paid intervention. And guess what happens next. The consultancy selling intervention produced the intervention-requiring result, from a subgroup of about 45 people, with no methodology. Have I mentioned the lack of methodology?

To make the point finer, it was the academics who reported the self-correction instead of a need for expensive consultants. The “evidence is piling up” story seems to totally ignore the fact that one story cancels the other one.

Right Forbes?