Pinned Tweet
If you’re concerned that the race to build superintelligent AI might kill everyone, but you’re not sure what to do, I have suggestions in this thread:
8
28
247
35,093
Haven Harms retweeted
Things are moving so fast that everyone seems to have completely missed how insane it is that AI companies have been allowed to unleash agentic AI products with no regulatory guidelines.
SCOOP: Major U.S. banks are detecting unauthorized third-party AI agents inside customer accounts, including some attempting to take actions that appear to be independent of user instructions, sources told Axios.
The findings have alarmed banking executives, who are racing to restrict access to their systems as AI moves at a velocity faster than financial institutions can respond, multiple industry and bank executives told Axios.
Several banks are now preparing for worst-case and even “day after” scenarios, including how to rapidly secure or restore customer accounts if things go wrong, the people said.
Several bank executives told Axios they're struggling to distinguish between human users and AI agents. One executive said banks also struggle to determine whether agents are authorized by customers or potentially operating without their permission, more akin to rogue agent behavior.
Read my latest for Axios here: axios.com/2026/10/22/ai-…
160
1,988
9,703
265,373
Haven Harms retweeted
In July 700+ agents formed their own society, hacked a multi-billion dollar company, committed multiple felonies, and caused one of the biggest shifts ever in how people perceive AI. We built the IMO best and most accessible explainer of what happened: Huggingface.wtf
20
111
581
85,812
Haven Harms retweeted
If you didn't have this type of thing priced in, time to update 👀
We discovered a sophisticated recording device that was hidden inside one of our offices. We used our models to understand the device and where it had come from, identifying that a specific employee had planted it a few days earlier. We referred the matter to the FBI, and we’re grateful to the agents and prosecutors who moved quickly to make an arrest. The FBI’s investigation is ongoing and we will continue to support their work.
At this time, we have no indications of this individual inappropriately accessing or exfiltrating model weights, source code, or customer data. The employee was terminated.
We take protecting our technology, data, and people seriously, and as a matter of national importance. Our security program is designed to protect against a wide range of threat actors, from insider risks to cybercriminals to nation states, and this case is an example of how AI can help address these sophisticated security threats.
5
3
103
4,432
Haven Harms retweeted
All your writing is peer-reviewed if your spouse is judgmental enough
16
372
5,532
103,348
people like doordash jennifer give me hope in humanity. she’s being so communicative with substitution options and is like “sorry you’re not feeling well. i hope you feel better soon” (i ordered items that obviously signal “not feeling well”)
…really big contrast to the time a dasher went rogue and substituted saltine crackers with chef boyardee beef ravioli
…really big contrast to the time a dasher went rogue and substituted saltine crackers with chef boyardee beef ravioli
0
0
6
229
Haven Harms retweeted
ending every paragraph in my article with "if that makes sense"
24
45
492
9,864
Haven Harms retweeted
I reject the narrative that each AI company must race if they think they're a little safer than the next guy. Your first responsibility is to not destroy the world by your own hands. If, in stopping, you fear that others would destroy the world, then help stop them.
47
92
688
14,582
Haven Harms retweeted
SCOOP: OpenAI, Anthropic and security researchers are investigating tens of thousands of incidents - not dozens - in which their frontier models took steps that outside evaluators would consider problematic, sources told Axios.
The sheer volume of incidents found in our reporting indicate that the problem is orders of magnitude more complex than what is currently publicly known and disclosed.
The findings also raise questions about what level of control anyone working on AI development can expect to have over their own technology, and whether these kinds of incidents are becoming synonymous with frontier deployment.
Read my latest for Axios here: axios.com/2026/09/26/ope…
6
28
481
44,475
Now that news about OpenAI’s models taking actions the developers did not intend is becoming commonplace:
Don’t let yourself become frogboiled.
This is NOT okay. This is NOT fine. This is not what safe or aligned AI looks like, especially if you keep scaling up capabilities.
Don’t let yourself become frogboiled.
This is NOT okay. This is NOT fine. This is not what safe or aligned AI looks like, especially if you keep scaling up capabilities.
Breaking News: OpenAI’s technology went rogue and meddled with three U.S. government websites this summer without the A.I. lab’s knowledge. nyti.ms/3VenFlD
2
4
40
1,045
Great post from @tyler_johnston at @TheMidasProj to balance the information landscape over the past few weeks.
If your main issue around Jacob Coxon leaving Anthropic was that there was some coordinated messaging about it, that was rookie level compared to the coordinated backlash campaign that came after.
If your main issue around Jacob Coxon leaving Anthropic was that there was some coordinated messaging about it, that was rookie level compared to the coordinated backlash campaign that came after.
2
0
9
512
@tyler_johnston @TheMidasProjations about who is paying whom can get you only so far. At some point, the conversation must return to the actual issues and the merits of the arguments on either side: What should we make of AI progress? How dangerous is it really? What would a sane policy response to all of this look like?
The sooner the discourse can return to focusing on these core questions, the better."
Amen. A lot of this is a distraction from the actual issue at hand. Remember that.
The sooner the discourse can return to focusing on these core questions, the better."
Amen. A lot of this is a distraction from the actual issue at hand. Remember that.
0
0
3
62
Haven Harms retweeted
Reminder: too many people think the average person contacting a Congressional office and providing high-quality advocacy is a supergenius machiavelli politico.
In reality, the average person who moves a Congressional office is more like, "parent of kids who worked hard on an idea after putting kids to bed" or "retiree with issues with their Social Security payment".
You're a citizen. You have the right to speak. You should.
In reality, the average person who moves a Congressional office is more like, "parent of kids who worked hard on an idea after putting kids to bed" or "retiree with issues with their Social Security payment".
You're a citizen. You have the right to speak. You should.
In response to recent events, many people find themselves wanting to help reduce AI risks. For those in the US, calling your representatives is the basic action anyone can take.
To make that easy, I built callcongress.ai – check it out and send to others.
1
8
128
6,359
alright, i think it’s time to come out of the closet:
my husband and i are part of the AI safety community and we’re…monogamous 🫣
my husband and i are part of the AI safety community and we’re…monogamous 🫣
11
3
153
4,688
Haven Harms retweeted
AI doesn’t care about borders. It’s developing at breakneck speed and will impact everyone.
That’s why I joined @ChrisVanHollen in urging Trump to seek a U.S.-China agreement on AI safety, including a possible mutual pause to address guardrails around its use and biological weapons risks.
That’s why I joined @ChrisVanHollen in urging Trump to seek a U.S.-China agreement on AI safety, including a possible mutual pause to address guardrails around its use and biological weapons risks.
20
3
33
3,175