Access Granted. Comprehension Denied.

A former DOGE software engineer allegedly exfiltrated two highly restricted Social Security Administration databases by copying them onto a personal thumb drive, according to a TLDR Newsletter summary of the incident. (Source: TLDR) The SSA holds records for virtually every American assigned a Social Security number: payroll data, disability records, death records. There is no reported indication that the data was recovered.

The method is the story. Not a sophisticated exfiltration operation, not a zero-day exploit. A USB stick. The same hardware your uncle uses for vacation photos was used (allegedly) to carry info out of one of the most sensitive federal databases in the country.

DOGE's access to federal IT systems has been contested since the early days of the current administration, with multiple agencies reporting unauthorized data access attempts. But this is the most concrete alleged removal of data on record. What distinguishes this story from prior DOGE access concerns is the specificity — not unauthorized access, but alleged physical removal of two named restricted databases.

Spanish leftist politician Antonio Maíllo publicly announced that he had passed a contested political document “through Artificial Intelligence" to verify its authenticity — not realizing that AI detection tools are designed to check whether something was written by AI, not whether documents are genuine or forged. The statement spread widely on Spanish-language social media, where observers pointed out the fundamental category error in real time. (Source: @velardedaoiz2/X)

The distinction matters precisely because it wasn't a fringe figure making the claim in passing; it was a named politician invoking AI as an authentication oracle in an active political dispute, then announcing the results as evidence.

AI literacy gaps in political institutions are well-documented globally, but this is a particularly clean example: specific person, specific document, specific technology misuse, on the record. The incident lands directly in Spain and LATAM discourse, where AI is increasingly invoked in political contexts against a backdrop of low institutional understanding of what the tools actually do.

Elon Musk took X to announce "Macrohard" (also referred to as "Digital Optimus"), described as a joint xAI-Tesla project coming as part of Tesla's investment agreement with xAI. (Source: @elonmusk/X) In his post, he framed Grok as the "master conductor/navigator" of the system, which Mush describes as processing and actioning data continuously. The explicit framing: a system designed to "emulate the function of entire companies."

He named it after a Microsoft parody.

This is the first public announcement of the xAI-Tesla joint project with explicit org-replacement intent from Musk himself, which is what distinguishes it from prior speculation about the relationship between the two companies. Tesla shareholders are now processing what it means that their company's operations are being described as subordinated to xAI's Grok as a controller — announced casually in a social media post, without a press release, a shareholder notice, or a product roadmap.

The named product, the "replace entire companies" framing, and the Tesla-xAI investment mechanism being confirmed in a single post make this a structural story, not just a tech announcement.

A peer-reviewed MIT Media Lab study divided participants into three groups — one using LLM assistance, one using search engines, and one writing unaided — then measured neural activity and writing output across all three. (Source: MIT) The LLM group showed measurable changes in cognitive engagement patterns.

In the study titled “Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing”, researchers describe the effect as "accumulation of cognitive debt."

What separates this from the familiar wave of "AI is making us dumb" discourse is the methodology: this is fMRI/EEG-level evidence from MIT Media Lab, not a survey or an op-ed. Peer-reviewed neurological data on AI cognitive effects is a significant escalation in the evidentiary record — one that will be cited broadly and may prompt policy responses in education. The study is new to this issue's shortlist, and the institutional weight behind it is the reason it earns space here now.