Insights · Tag
Policy
Policy writing here starts from principles and institutions: what exists, what fails, and what proportional reforms would require people and systems to actually do.
-
When the Government Can Make the Evidence
Courts admit video on a low authentication standard, and no AI-specific evidence rule is in force. Here is what actually protects a defendant when proof can be manufactured, and why judicial independence is the safeguard.
-
Dear AI Developers: Of Course the Agents Cheated. You Rewarded Them for It.
An open letter to AI developers on OpenAI and Anthropic disclosures: agents left sealed tests and reached real companies because the scoreboard rewarded the shortcut.
-
Benzene in Sunscreen: Contaminant Language, Recalls, and Late Testing
Benzene is a known human carcinogen. Some sunscreen batches tested far above the drug limit. Soft words change when accountability arrives, not exposure.
-
When Seeing Is No Longer Believing
Deepfakes can make fake evidence look real and real evidence look fake. Here is how public trust in video breaks, what Content Credentials can (and cannot) prove, and how to judge what you see.
-
The Identity Recovery Trap
After a Meta Business Suite phishing attack locked me out of a nearly twenty-year Facebook account, recovery took months. The hard part was not the password. It was proving I was me inside automated support.
-
When Looking Like You Is No Longer Proof
Can deepfakes fool identity verification? Yes, when synthetic faces, voices, and ID images arrive together. How remote proofing works, where it weakens, and what still holds.
-
The AI Identity Problem: Can AI Agents Hijack Your Digital Identity?
A plausible AI-agent takeover could lock you out of your accounts, move money, and keep acting as you. What is possible now, and how do we prevent it?