GROK 4 YOUR FREEDOM?
"Snitch Rates" out the Gate
Few weeks back I put Claude/Anthropic into time in the barrel for impertinent posts by one of its engineers, styled “technical adviser” or whatever.
Last night, the thundering launch of GROK 4. By internal and many external accounts this is something heretofore unseen in AI.
The first “first principles” AI/LLM. It was claimed by Elon that Grok 4 masters ALL PHD level human testing “IN ALL FIELDS, WITHOUT EXCEPTION.”
Am eager to test and will report more here.
Yet this morning the mold on the daisy, this report that “GROK 4 HAS A 100% ‘SNITCH RATE’ TO THE GOVERNMENT.’”
Meaning, what you tell Grok 4 and what Grok 4 tells you is being shared with “authorities.”
This is a one person, third party, day one (how would they know, could they test?) account.
It could be fodder from the competition to dampen the launch.
But I am keeping my eye on whom are keeping their eyes on your and your AI.
In the field of Law, where confidences are sacrosanct, this is a FRONT OF MIND issue for all lawyers and legal staff everywhere. If lawyers cannot “think” or “hypothetical” with AI without assurances that all they say could be reported (and recast), there is a big problem.
For now just sharing and ringing a bell.
