Anthropic Drafts Its Own AI Safety Bill
Elon Musk Podcast
October 19, 2026
The artificial intelligence firm Anthropic has privately shared a draft AI safety bill with United States lawmakers to help shape upcoming federal regulation.
TopicsTechnology
SPEAKER_1 (0:01)
When you need to build up your team to handle the growing chaos at work, use Indeed Sponsored Jobs. It gives your job post the boost it needs to be seen and helps reach people with the right skills, certifications and more. Spend less time searching and more time actually interviewing candidates who check all your boxes. Listeners of this show will get a $75 sponsored job credit at indeed.com/podcast. That's indeed.com/podcast. Terms and conditions apply. Need a hiring hero? This is a job for indeed sponsored jobs.
SPEAKER_2 (0:30)
Chronic migraine is 15 or more headache days a month, each lasting four hours or more.
SPEAKER_3 (0:35)
Botox, on about your line, and toxin A prevents headaches in adults with chronic migraine before they start. It's not for those with 14 or fewer headache days a month. It prevents on average eight to nine headache days a month versus six to seven for placebo.
SPEAKER_4 (0:48)
Prescription Botox is injected by your doctor. Effects of Botox may spread hours to weeks after injection, causing serious symptoms. Alert your doctor right away as difficulty in swallowing, speaking, breathing, eye problems or muscle weakness can be signs of a life-threatening condition. Patients with these conditions before injection are at highest risk. Side effects may include allergic reactions, neck and injection site pain, fatigue and headache. Allergic reactions can include rash, welts, asthma symptoms and dizziness. Don't receive Botox if there's a skin infection. Tell your doctor your medical history, muscle or nerve conditions, including ALS, Lou Gehrig's disease, Myasthenia Gravis or Lambert-Eton syndrome and medications, including botulinum toxins, as these may increase the risk of serious side effects.
SPEAKER_2 (1:24)
Why wait? Ask your doctor. Visit botoxchronicmigraine.com or call 1-800-44-BOTOX to learn more.
SPEAKER_5 (1:30)
Is your home wishlist just sitting in your cart? Well, get ready to hit checkout because Way Day at Wayfair starts October 24th with up to 80% off everything home. That new sofa, the rug you've been eyeing, the upgrades you've been putting off, they could all be way less during Way Day.
It's Wayfair's biggest sale of the year. So make your list and mark your calendar. Shop Way Day starting October 24th for up to 80% off everything home at wayfair.com.
SPEAKER_4 (1:57)
Wayfair, every style, every home.
SPEAKER_6 (2:00)
Anthropic quietly pitched a draft of its own AI safety bill to US lawmakers.
SPEAKER_7 (2:06)
Yeah, Reuters and a company spokesperson essentially confirmed this. The draft builds directly on Anthropic's own internal advanced AI framework, though they are effectively taking their company policy and trying to make it federal law.
SPEAKER_6 (2:19)
Which, you know, sets up an immediate tension. You have a major tech company writing the rules meant to govern its own products and handing that text directly to the people in charge of regulating them.
SPEAKER_7 (2:29)
Right. It forces you to ask, when a frontier labrates the laws, are they offering the most informed protection possible or simply defining the rules so they always win?
SPEAKER_6 (2:40)
Well, the document itself tries to answer that by establishing very concrete thresholds. The draft requires a specific trigger before the government is allowed to step in and block a model from being released.
SPEAKER_7 (2:51)
Those triggers are very specific, right?
SPEAKER_6 (2:52)
Extremely. It requires an imminent risk of more than $1 billion in damage or killing or seriously injuring more than 50 people.
SPEAKER_7 (3:00)
But quantifying a theoretical $1 billion in damage before a model is even released is a bizarre mechanical problem. How does an independent evaluator look at code sitting on a server and project a highly specific financial catastrophe?
SPEAKER_6 (3:16)
It is exactly like asking an accounting department to run a pre-crime division.
SPEAKER_7 (3:20)
It really is.
You have evaluators trying to build financial models for disasters that haven't even happened yet. Imagine you are the auditor and you have to prove to a legal standard that this exact software will cause a 10-figure financial loss.
SPEAKER_6 (3:35)
Right, not $900 million, $1 billion.
SPEAKER_7 (3:38)
Exactly. And the mechanics of proving that are nearly impossible.
If a model has a flaw that might cause a cascading failure in algorithmic trading, you can't guarantee a specific dollar amount until the market actually reacts.
SPEAKER_6 (3:52)
By setting the threshold at an arbitrary billion dollars, the draft essentially guarantees that no evaluator will ever be confident enough to pull the alarm. Though it does limit arbitrary government intervention, which is probably the goal.
SPEAKER_7 (4:06)
Sure, but it sets a ceiling so high that everyday issues are completely shielded from this specific shutdown mechanism.
SPEAKER_6 (4:13)
Yeah, things like biased algorithms, copyright disputes or localized data privacy violations, they don't even come close to those numbers.
18 more minutes of transcript below
Thousands of transcripts fetched by people building searchable podcast archives
Fetch the whole transcript
The demo key returns a sample episode in full, no card needed:
curl -H "x-api-key: pt_demo" \
https://spoken.md/transcripts/1000651996090Markdown with the speakers named, for your notes, your knowledge base, or anything that makes HTTP calls.
From $0.10 per transcript. No subscription. Credits never expire. Prices exclude VAT, added at checkout for EU customers. Not what you expected? Email us within 14 days with 20 or fewer credits used and we refund the pack in full.
Using your own key:
curl -H "x-api-key: YOUR_KEY" \
https://spoken.md/transcripts/1000797039046