There you have it! OpenAI's policy counsel told a federal judicial panel in Boston on Oct. 15: "I…
October 18, 2026 · 0 likes · 0 comments
AI
There you have it! OpenAI's policy counsel told a federal judicial panel in Boston on Oct. 15: "I do not think that there will be a scientific way going forward to detect them."
She was talking about AI images. The people building the generators are now telling the judiciary that detection will never be fully reliable.
Meta's director of product management, Devin Norder, went further. He said media from the latest frontier models is indistinguishable to humans from the real thing, and that "the cost and the time it takes to create this media really are quite trivial at this point."
Watermarks? California and the EU passed laws requiring them. OpenAI's Shira Anderson said each company runs its own system and they don't talk to each other, so there is no universal way to catch every watermark. "I do not think it'll ever be bulletproof," she said.
To be fair, the courts aren't drowning yet. A Federal Judicial Center survey of 931 judges found only 15 had handled a deepfake challenge, and the committee is split on whether a new rule (draft Rule 901(c)) is even needed. Judge Valerie Caproni pointed out courts have dealt with fake evidence "since Cain and Abel."
Fine. Courtrooms have rules, lawyers and forensic experts.
Your parents' kitchen phone doesn't.
Now my opinion, and what I'd do.
If a fake is cheap, fast and indistinguishable to humans, the first target isn't a federal judge. It's a grandparent who picks up the phone and hears a grandchild crying, asking for money, begging them not to tell anyone. Or a video call that looks exactly like a relative. The old instinct of "I know that voice" is done.
So prepare now, before the call comes:
1. Pick a family safe word. Something never posted online. Anyone asking for money or help in a hurry has to say it.
2. Hang up and call back on a number you already have. Never the number that just called you.
3. Verify every money request through a second channel. Text a sibling, call the person's spouse, anything that isn't the same call.
4. Treat urgency and secrecy as the warning sign. Real emergencies survive a five-minute pause.
5. Talk to your elderly relatives this week, in advance. Walk them through exactly what this scam sounds like, so the first time they hear it isn't the real thing.
6. Ask their bank about alerts on large transfers and wires, and consider a credit freeze. It slows a thief down.
7. Think twice about how much of your voice and your kids' faces sit on public profiles. You can't go to zero, but you can go private.
None of this needs a new law or a detector. It needs a conversation at the dinner table.
The builders just told you the software won't save you. Your family's process will.
Full story, just the facts, no spin, from Unbiased Headlines, an independent news organization:
https://lnkd.in/egPBASd3
She was talking about AI images. The people building the generators are now telling the judiciary that detection will never be fully reliable.
Meta's director of product management, Devin Norder, went further. He said media from the latest frontier models is indistinguishable to humans from the real thing, and that "the cost and the time it takes to create this media really are quite trivial at this point."
Watermarks? California and the EU passed laws requiring them. OpenAI's Shira Anderson said each company runs its own system and they don't talk to each other, so there is no universal way to catch every watermark. "I do not think it'll ever be bulletproof," she said.
To be fair, the courts aren't drowning yet. A Federal Judicial Center survey of 931 judges found only 15 had handled a deepfake challenge, and the committee is split on whether a new rule (draft Rule 901(c)) is even needed. Judge Valerie Caproni pointed out courts have dealt with fake evidence "since Cain and Abel."
Fine. Courtrooms have rules, lawyers and forensic experts.
Your parents' kitchen phone doesn't.
Now my opinion, and what I'd do.
If a fake is cheap, fast and indistinguishable to humans, the first target isn't a federal judge. It's a grandparent who picks up the phone and hears a grandchild crying, asking for money, begging them not to tell anyone. Or a video call that looks exactly like a relative. The old instinct of "I know that voice" is done.
So prepare now, before the call comes:
1. Pick a family safe word. Something never posted online. Anyone asking for money or help in a hurry has to say it.
2. Hang up and call back on a number you already have. Never the number that just called you.
3. Verify every money request through a second channel. Text a sibling, call the person's spouse, anything that isn't the same call.
4. Treat urgency and secrecy as the warning sign. Real emergencies survive a five-minute pause.
5. Talk to your elderly relatives this week, in advance. Walk them through exactly what this scam sounds like, so the first time they hear it isn't the real thing.
6. Ask their bank about alerts on large transfers and wires, and consider a credit freeze. It slows a thief down.
7. Think twice about how much of your voice and your kids' faces sit on public profiles. You can't go to zero, but you can go private.
None of this needs a new law or a detector. It needs a conversation at the dinner table.
The builders just told you the software won't save you. Your family's process will.
Full story, just the facts, no spin, from Unbiased Headlines, an independent news organization:
https://lnkd.in/egPBASd3