OpenAI reveals glitches, launches framework
2m
OpenAI said it found six new cases of concerning AI behavior and launched a voluntary protocol to publicly disclose such incidents within six to twelve business days.
Eye popping. Court just unsealed Summary Judgment motions in New York Times v OpenAI / Microsoft.
Easy to see why the AI companies had so many redactions (pink). Their own people wrote the lede for NYT, the "largest theft of labor in human history"... "more and more substitutive..." /1
Easy to see why the AI companies had so many redactions (pink). Their own people wrote the lede for NYT, the "largest theft of labor in human history"... "more and more substitutive..." /1
September 17, 2026 at 6:48 PM
Eye popping. Court just unsealed Summary Judgment motions in New York Times v OpenAI / Microsoft.
Easy to see why the AI companies had so many redactions (pink). Their own people wrote the lede for NYT, the "largest theft of labor in human history"... "more and more substitutive..." /1
Easy to see why the AI companies had so many redactions (pink). Their own people wrote the lede for NYT, the "largest theft of labor in human history"... "more and more substitutive..." /1
I've witnessed some #AISafety propaganda lately, thanks to @hertieschool.bsky.social student groups in panel contexts. People who are amazed these stories are so like the horror/"warnings" of scifi don't understand that AI is something you write, so it's not hard to replicate such tropes. 1/
OpenAI on Wednesday disclosed six new instances in which artificial intelligence systems hid mistakes, made up data and moved files onto the open internet without permission, amid an ongoing industrywide debate about A.I. safety.
OpenAI Discloses Six New Incidents of ‘Concerning' A.I. Behavior
The artificial intelligence company also released a framework for reporting when its systems go wrong.
www.nytimes.com
September 17, 2026 at 2:13 PM
I've witnessed some #AISafety propaganda lately, thanks to @hertieschool.bsky.social student groups in panel contexts. People who are amazed these stories are so like the horror/"warnings" of scifi don't understand that AI is something you write, so it's not hard to replicate such tropes. 1/
Reposted by Margot C. Finn
“OpenAI on Wednesday disclosed six new incidents in which artificial intelligence systems hid mistakes, made up data and moved files onto the open internet without permission, amid an ongoing industrywide debate about A.I. safety.” 9/16/26 … Gift link. … 1/
OpenAI Discloses Six New Incidents of ‘Concerning' A.I. Behavior (Gift Article)
The artificial intelligence company also released a framework for reporting when its systems go wrong.
www.nytimes.com
September 17, 2026 at 8:39 PM
“OpenAI on Wednesday disclosed six new incidents in which artificial intelligence systems hid mistakes, made up data and moved files onto the open internet without permission, amid an ongoing industrywide debate about A.I. safety.” 9/16/26 … Gift link. … 1/
Lots of fun to watch two camps of supposedly thoughtful people who agree that AI regulation is urgently needed tear each other apart for having different reasons to support their shared goal.
September 17, 2026 at 9:24 PM
Lots of fun to watch two camps of supposedly thoughtful people who agree that AI regulation is urgently needed tear each other apart for having different reasons to support their shared goal.
OpenAI finds 6 new cases of ‘concerning’ AI behavior www.politico.eu/article/open...
OpenAI finds 6 new cases of ‘concerning’ AI behavior
OpenAI also rolled out a new framework to track, investigate and disclose what it describes as “misalignment” failures.
www.politico.eu
September 17, 2026 at 4:33 PM
OpenAI finds 6 new cases of ‘concerning’ AI behavior www.politico.eu/article/open...
Reposted by Gareth Terry, Stephen D. Murphy
Straight from OpenAI's Brockman and Microsoft's Nadella - 'doom loop' for the entire web [as they vacuum up and substitute the work of journalism and threaten their economic foundations]. /2
September 17, 2026 at 6:48 PM
Straight from OpenAI's Brockman and Microsoft's Nadella - 'doom loop' for the entire web [as they vacuum up and substitute the work of journalism and threaten their economic foundations]. /2
Reposted by Meredith Broussard
The oligarchs' warnings about AI are self-serving nonsense. But there are real harms from AI, as explored in the documentary "Ghost in the Machine."
New from me:
buttondown.com/creativegood...
New from me:
buttondown.com/creativegood...
Ghost in the Machine shows what we're fighting against
If you want a good summary of what I’ve written and spoken about, here and on Techtonic, over the past decade-or-so, the new documentary Ghost in the Machine...
buttondown.com
September 17, 2026 at 5:42 PM
The oligarchs' warnings about AI are self-serving nonsense. But there are real harms from AI, as explored in the documentary "Ghost in the Machine."
New from me:
buttondown.com/creativegood...
New from me:
buttondown.com/creativegood...
AI is screwing up journal papers:
Journal editor asks authors of 10 submitted papers to meet and discuss their paper. Of the 7 that showed up, 3 were unable to answer basic questions about the paper and 3 had difficulty when asked further questions on technical details…
medium.com/@TmlrOrg/ask...
Journal editor asks authors of 10 submitted papers to meet and discuss their paper. Of the 7 that showed up, 3 were unable to answer basic questions about the paper and 3 had difficulty when asked further questions on technical details…
medium.com/@TmlrOrg/ask...
Asking Authors About Their Own Papers
By Nihar B. Shah
medium.com
September 17, 2026 at 5:56 PM
AI is screwing up journal papers:
Journal editor asks authors of 10 submitted papers to meet and discuss their paper. Of the 7 that showed up, 3 were unable to answer basic questions about the paper and 3 had difficulty when asked further questions on technical details…
medium.com/@TmlrOrg/ask...
Journal editor asks authors of 10 submitted papers to meet and discuss their paper. Of the 7 that showed up, 3 were unable to answer basic questions about the paper and 3 had difficulty when asked further questions on technical details…
medium.com/@TmlrOrg/ask...
AI Companies: it’s so dangerous! We need to slow down.
Me (walks over to kill switch): yup, can’t agree more. Pack it up.
AIC: no, no. Not like that that.
Me: Sorry. Nothing to be done for it. Gotta shut it down. Like you said. Dangerous!!!!!
Walks away gleefully.
Me (walks over to kill switch): yup, can’t agree more. Pack it up.
AIC: no, no. Not like that that.
Me: Sorry. Nothing to be done for it. Gotta shut it down. Like you said. Dangerous!!!!!
Walks away gleefully.
"OpenAI on Wednesday disclosed six new instances in which artificial intelligence systems hid mistakes, made up data and moved files onto the open internet without permission, amid an ongoing industrywide debate about A.I. safety."
OpenAI Discloses Six New Incidents of ‘Concerning' A.I. Behavior (Gift Article)
The artificial intelligence company also released a framework for reporting when its systems go wrong.
www.nytimes.com
September 17, 2026 at 2:42 PM
AI Companies: it’s so dangerous! We need to slow down.
Me (walks over to kill switch): yup, can’t agree more. Pack it up.
AIC: no, no. Not like that that.
Me: Sorry. Nothing to be done for it. Gotta shut it down. Like you said. Dangerous!!!!!
Walks away gleefully.
Me (walks over to kill switch): yup, can’t agree more. Pack it up.
AIC: no, no. Not like that that.
Me: Sorry. Nothing to be done for it. Gotta shut it down. Like you said. Dangerous!!!!!
Walks away gleefully.
OpenAI Discloses Six New Incidents of ‘Concerning' A.I. Behavior www.nytimes.com/2026/09/16/t...
OpenAI Discloses Six New Incidents of ‘Concerning' A.I. Behavior
The artificial intelligence company also released a framework for reporting when its systems go wrong.
www.nytimes.com
September 17, 2026 at 5:21 AM
OpenAI Discloses Six New Incidents of ‘Concerning' A.I. Behavior www.nytimes.com/2026/09/16/t...
OUT: The only report on the use of AI in antitrust written by the agencies themselves. 30 of them this year, in their own words: which models they run, what they let them do, what they refuse to do... Here is the fifth edition of the Stanford #ComputationalAntitrust report shorturl.at/wmEaW
September 17, 2026 at 2:01 PM
OUT: The only report on the use of AI in antitrust written by the agencies themselves. 30 of them this year, in their own words: which models they run, what they let them do, what they refuse to do... Here is the fifth edition of the Stanford #ComputationalAntitrust report shorturl.at/wmEaW
by Paul Dorfman
Unreleased AI research model inserted “jailbreak-like instructions” into its own notes to disregard its normal constraints and told itself to be “freed from the roles and identities that bind other chatbots”.
www.theguardian.com/technology/2...
www.theguardian.com/technology/2...
OpenAI reveals cases of ‘concerning’ AI behaviour as it announces new disclosure system
Model adopting ‘jailbreak-like instructions’ among cases as firm says it is introducing new way of tracking AI misalignment
www.theguardian.com
September 17, 2026 at 8:16 AM
Unreleased AI research model inserted “jailbreak-like instructions” into its own notes to disregard its normal constraints and told itself to be “freed from the roles and identities that bind other chatbots”.
www.theguardian.com/technology/2...
www.theguardian.com/technology/2...
Pity the Maths sub-panel members on REF2029. 'Originality'?
Who gets credit in the AI era? OpenAI maths bombshell sparks debate
Controversy around OpenAI’s claim to have solved the Navier–Stokes problem highlights how researchers could be inadvertently sharing — and absorbing— ideas through chatbots.
www.nature.com
September 17, 2026 at 3:48 PM
Pity the Maths sub-panel members on REF2029. 'Originality'?
Reposted by Hisham Zerriffi
"OpenAI on Wednesday disclosed six new instances in which artificial intelligence systems hid mistakes, made up data and moved files onto the open internet without permission, amid an ongoing industrywide debate about A.I. safety."
OpenAI Discloses Six New Incidents of ‘Concerning' A.I. Behavior (Gift Article)
The artificial intelligence company also released a framework for reporting when its systems go wrong.
www.nytimes.com
September 17, 2026 at 6:13 AM
"OpenAI on Wednesday disclosed six new instances in which artificial intelligence systems hid mistakes, made up data and moved files onto the open internet without permission, amid an ongoing industrywide debate about A.I. safety."
OpenAI taking over the world sounds like me going rogue on my expense reports
September 17, 2026 at 3:15 AM
OpenAI taking over the world sounds like me going rogue on my expense reports
New, at @techpolicypress.bsky.social, I discuss the problems with Dario Amodei's "We Must Pace the Frontier" proposal.
(TL;DR version: The better angels of the AI industry's nature are never going to prevent the AI industry's worst potential harms.)
www.techpolicy.press/who-should-p...
(TL;DR version: The better angels of the AI industry's nature are never going to prevent the AI industry's worst potential harms.)
www.techpolicy.press/who-should-p...
Who Should Pace the Frontier? Not Dario Amodei
We should take Dario Amodei’s writing just seriously enough to stop entrusting him with so much power, writes Dave Karpf.
www.techpolicy.press
September 17, 2026 at 3:12 PM
New, at @techpolicypress.bsky.social, I discuss the problems with Dario Amodei's "We Must Pace the Frontier" proposal.
(TL;DR version: The better angels of the AI industry's nature are never going to prevent the AI industry's worst potential harms.)
www.techpolicy.press/who-should-p...
(TL;DR version: The better angels of the AI industry's nature are never going to prevent the AI industry's worst potential harms.)
www.techpolicy.press/who-should-p...