I’ve had a number of conversations with folks inside and outside government about the current situation with Anthropic, and here is what I believe to be true:
— As we know, Anthropic publicly released its Mythos class models earlier this week under the commercial name Fable.
— Fable is Mythos with guardrails. But if those guardrails fail, then you’ve exposed Mythos and its advanced cyber capabilities to people who shouldn’t have them. (Keep in mind that Anthropic itself widely promoted the idea that Mythos was a cyberweapon and needed to be regulated as such. They asked for government regulation of Mythos and championed the guardrails on Fable. If there is a vulnerability — big or small — it is Anthropic’s responsibility to patch.)
— A highly credible trusted partner of both Anthropic and the USG who was testing Fable came forward with a jailbreak of those guardrails. The Admin asked Dario to fix the jailbreak or de-deploy the model. Dario refused.
— In their blog post, Anthropic defended its decision by saying the jailbreak isn’t serious. That is not what the trusted partner and the USG believe; nor is that kind of minimizing language consistent with Anthropic’s brand as the AI safety company. It’s difficult to fathom how they could claim a jailbreak allowing operability of a cyber weapon could be defined as not “serious.”
— In the past, Anthropic has always said that safety must be top priority and taken super seriously. In this case, Anthropic prioritized the continued offering of the consumer model over safety.
— In reaction, the Admin issued the export control. The Admin did this reluctantly. It’s been very surprised that Anthropic hasn’t wanted to cooperate with a reasonable safety request (ie fixing the jailbreak issue). Anthropic’s reaction is very much at odds with their branding and ethos as a safe AI research community.
— The Admin’s hope now is that Anthropic remediates the safety issue, the export control is lifted, and Fable goes back into general release. The Admin wants all of this to happen as soon as possible. It is frankly bewildered that Anthropic hasn’t wanted to comply with safety requests that it previously said were its highest priority.
— Those trying to misdirect and tie this action to the prior DoW/Anthropic issues are wrong. The Admin values Anthropic’s technical capabilities and feels that this issue, while serious, should be easily resolved. The ball is in Anthropic’s court.
.@BernieSanders , it is a time to celebrate. @elonmusk has created enormous value for society by building @SpaceX, driving down the cost of rocket launches and creating a global satellite communication network that has brought high speed, low-cost internet and communication access to hundreds of millions and eventually billions of people along with critical advantages for our military and our nation’s defense.
SpaceX and its technologies will cause an acceleration in the growth of wages and wealth creation globally, including in some of the poorest communities in the U.S. and around the world.
Access to low-cost, high speed communications everywhere will allow children around the world to be educated, families to build businesses, and life-saving medical knowledge and care to be available everywhere.
SpaceX will materially bring down the cost of compute, advancing AI and humanity.
Meanwhile, 4,000 SpaceX employees yesterday became millionaires, including hourly wage employees who you claim you are trying to help.
The Elon Musks of the world drive growth, global GDP, and provide access to goods and services at lower cost that would otherwise not exist.
Elon’s nominal trillionaire status is due to his ownership of SpaceX, Tesla, Neuralink, the Boring Company and his other initiatives that have brought new technologies that improve our everyday lives.
Elon is not sitting on a trillion dollar pile of cash, jewelry and gold. He is using his controlling stakes in his companies to advance mankind. Elon’s companies don’t pay dividends. They reinvest all of their capital to accelerate innovation and value creation.
Elon is working 24/7 for all of us. He deserves respect and appreciation, not smears.
Bernie, your socialism would never allow a SpaceX to be built. Socialism has only proven to impoverish mankind and lead to death and destruction.
We need to create the conditions for more SpaceXs to be built, not attack the great entrepreneurs who are helping to advance our country.
The richest guy on Earth SHOULD be the guy making cutting edge cars and rockets instead of dudes who sell purses and perfumes or dudes who run investment firms or dudes who made Facebook.
One piece of advice we got during YC was to explain our company using verbs instead of nouns.
Early on, I walked into a meeting and did the opposite:
“We’re building a cloud platform for AI”
No one knew that that meant, their eyes glazed over. Then I started saying this instead:
“We containerize your code and run it on GPUs in the cloud so you don’t have to manage the infra yourself”
That clicked way more. Our brains understand verbs because they’re more concrete. If you describe your company using nouns, you risk people not understanding you.
And no one buys or invests in things they don’t understand.
This shows yet again how this limitation was never about "safety" but about Anthropic doing stuff just because they thought they can.
I am increasingly sceptical Anthropic really cares about safety, and not just their business interests (limiting competition where they can)
@GergelyOrosz hi welcome to my long running conspiracy theory that Dario is actually the Villain and he wants to control us all and be our Dad and protect us at all costs even if that means destroying our lives.
Sheryl. Your article exemplifies the biased reporting we have come to expect from you and @nytimes. It was unfair, inimical, and inaccurate. All one needs to refute your argument is to glance at my publicly available calendar and to review my unprecedented list of accomplishments on a wide range of issues, all of which I drove. You evidently never undertook these foundational due diligences. Why let facts obscure a good story?
You fault me for missing a couple of monthly counselor meetings. However, I meet one-on-one with my counselors every day to decide policy and strategy. We schedule the monthly meetings to give the divisions a chance to keep each other informed about HHS-wide policies with which I’m already intimately familiar. Had you read my calendar, you would have seen that I have back-to-back meetings all day, every day, with both career and political staff, with my counselors and with outside stakeholders, interspersed with press conferences and other policy announcements.
I am knowledgeable and active on every issue in every division of my department, and I always make the final decisions. I meet with the principals at FDA, NIH, CDC, and my senior counselor every morning, something, I’m told, is unprecedented in HHS history. I try to get out of the office between 4:30 and 6:00 PM, so that I can spend three hours, in quiet, responding to emails. I normally work until 11 PM every night, mostly on phone calls to staff.
In order to prove your preconceived case for my disengagement, you quote anonymous employees, some of whom I fired or who quit to avoid being fired. You also deceptively quote HHS employees without identifying whether they were among those I fired, thereby depriving your readers of the opportunity to make an independent judgment about their credibility.
I came into this job to change the culture of a broken agency that has presided over the worst decline in public health in American history. Of course I fired people—lots of them! It's an easy task for even the laziest journalist, to comb that flotsam and jetsam for malevolence toward the Trump administration. And of course, this species of journalist will always be able to find disgruntled individuals among the 70,000 employees of the Department from whom to cherry pick "facts" to flesh out a preordained hit piece. All that is required for this brand of journalism is the ethical elasticity that you seem to have in spades. You had a preconceived thesis, and you set out to prove it. This is a widely accepted technique in journalism today, but I grew up in an era when it would not have been tolerated by the New York Times.
Ultimately, God puts us all on this earth to search for existential truths. I've tried to instill this mission at HHS by implementing gold standard research to end the regime of politicized science that COVID exposed to the American public. There was a time that journalists were proud to be the fearless and uncompromising champions of truth. Standards have devolved, and journalism is dead. The Times now employs propagandists. Your capitulation to partisanship further compounds your journalistic challenges; since we all are aware of your predictable bias, we at HHS are unwilling to talk to you about the topics that are important. The fact that you have minimal access to decision makers leaves you covering trivia and relying on your own capacity for invention.
Btw. When I took this job, the building was empty. About 90% of the employees were not coming to work. I changed that, but your newspaper never covers my reforms. Nor did you cover the fact that my predecessor almost never showed up for work here during his four years in office. When we came in, there were still artifacts from the first Trump administration in many of our office drawers because no one showed up for work during the Biden years. Just as Rochelle Walensky spent her entire term as CDC Director in Cambridge, Xavier Becerra reportedly spent most of his term as HHS Secretary in California. (I live in California, but I’ve only been there once in fifteen months).
His only notable accomplishments here were losing 300,000 children, referred to HHS for custody and care, to human traffickers and drug runners, encouraging transgender surgeries, and disabling the entire program-integrity apparatus, allowing hundreds of billions of dollars of theft from my agency. I have set out to find the children Becerra lost. He is now the front-runner for the governor of California. These are not invented stories; they are genuine scandals that the Times will never cover, presumably, because the malefactors are Democrats.
Finally, you criticize me for spending time with the Indian tribes in Alaska. I consider that part of my job. I run the Indian Health Services, and I’ve had unprecedented success in transforming IHS from a backwater to a top priority for this department. I’ve made more trips to Indian country and to Indian health clinics and hospitals than any HHS secretary in history, and I’ve brought Indians into high positions on the sixth floor for the first time in agency history. This is another success story that the Times will never cover.
Brilliant idea! Next up: Apple randomly reboots your Mac if you're building competing tech, Gmail silently edits your email if you mention rival platforms, and Tesla Autopilot swerves if it detects you're working on self-driving cars.
All in the name of safety, of course. Because malicious actors controlling the world’s operating systems, inboxes and cars would be extremely dangerous!
SITUATION DETECTED: Microsoft is limiting internal employee use of Fable 5 over Anthropic's new data retention requirements, per The Verge.
Fable 5 requires data retention to operate its safety classifiers, unlike other Claude models which run under Zero Data Retention rules.
I will be honest:
1. Since Opus 4.7 I get pushback in unexpected ways from Anthropic models
2. GPT-5.5 doesn't do this
When I know what I want to do, I prefer #2. I don't like a vendor knowing better what I do, how I do it, with zero transparency. I humor Opus but like GPT...
NEW: malware developers added nuclear & biological weapons text to to their spyware.
Goal? To trigger LLM safety refusals... so that their spyware wouldn't be analyzed by an AI security scanner.
Cleanest practical example I can think of for why over-indexing on first order safety alignment is risky.
When closed (and open) models ship with aggressive refusals, they will be sprinkled with second-order blindspots that attackers will discover...and exploit.
We are only in the earliest days of attackers leveraging these features, and it wouldn't surprise me if users systems that need to handle complex cybersecurity issues demand that models be less safety-blunted.
In the weeds: @SocketSecurity's post also shows why intention matters in how you design a malware analysis pipeline to avoid prompt manipulation.
H/T to colleagues that shared this with me https://t.co/f3Aj9TYxU4