Ensure Legality of Agents @ https://xitter.cf/t.co/jtre7FLbdc US/India
San Francisco, US
Joined November 2008
- Tweets2.1K
- Following723
- Followers594
- Likes513
Pinned Tweet
Glad to announce partnership with @CarbonArcAI to bring global regulatory risk intelligence to wall-street and beyond. Carbon Arc is building the next generation data infrastructure and marketplace for fast moving applications without the legal, engineering, and process overheads. We are glad to be adding the regulatory coverage for these applications.
Carver Agents tracks regulatory notifications from 1,000+ regulators across 50+ countries, spanning finance, AI, data, energy, healthcare and more. We ingest data from official government bodies including press releases and enforcement actions, then turn them structured machine-consumable data: source and jurisdiction, key dates, named entities, the functions and industries impacted, the actions they trigger, and more.
For teams whose risk crosses borders, that structured feed becomes real signal:
• Trade surveillance teams can spot jurisdictions where enforcement is accelerating before it shows up in a filing
• Supply chain teams can see where data, energy or environmental rules are tightening before they disrupt a build
• M&A teams can screen cross-border targets for regulatory exposure they'd otherwise catch too late
• PE and project teams can track how regulatory pressure is shifting in a sector before it changes the math on a deal
If you're managing cross-border risk, in trade, supply chains, M&A or project finance, this is a new lens worth having. Reach out to the fantastic team at @CarbonArcAI or us to discuss how to incorporate it into your workflows.
Essentially the US and China are playing a high-risk high-reward game - imperial overstretch vs imperial technology.
“Whoever produces and controls frontier models acquires global leverage over innovation, production, science, culture, and, ultimately, advanced thought.” Read @nfergus, Lisa Klaassen, and @ericschmidt on the imperial nature of AI:
foreignaffairs.com/china/emp…
Great explanation of what we are seeing and likely to see on AI rollout front. Two key ideas I picked up were (a) only when we try to automate we will discover that there were hidden dependencies and bottlenecks. Addressing them will take time. (b) Processes have to be have redesigned to with the new tech available. This will also take time and we cant mix and match old process with new tech.
I think there will be legal software factory as well to deal with adversarial legal attacks
Just spent the last 7 days on the front lines of the vulnpocalypse fending off an active cyber attack on one of our products.
Some thoughts:
- Software factories are going to be required equipment for everyone. There’s just no way to keep up with the pace of highly automated adversaries without one.
- We were lucky because we had spent the last few months getting our software factory running well. It is nice to have for product development but proved to be absolutely essential this week.
- At the minimum you need to have a coding server in the cloud pre-provisioned with read only access to all your logs and all the services you depend on for observability. With that in place you are a prompt away from reasonably good situational awareness, ongoing monitoring with loops, and ability for coding agents to validate fixes against your actual, observed attack patterns.
- Shared coding server also unlocks multiplayer incident response where everyone can see every session and jump in as needed. Sharing session links that can be forked is much higher bandwidth than re-explaining things.
- You don’t need to be Ramp/Stripe/Uber scale to have a working software factory in the cloud. We are a team of 3. Ours is built with an @opencode server (thanks @thdxr!), a forked version of compound engineering optimized for automated workflows (thanks @kieranklaassen!) and a multiplayer agent in slack that handles crons (we have our own harness but pick your favorite here).
- To get our software factory up and running well, we had to stop all new product development for ~6 weeks and just focus on getting the validation systems working that enabled us to reliably shipping PR without human code review. This felt very painful in the moment but was totally worth the eventual speedup in velocity.
- KYC for @AnthropicAI & @OpenAI cyber programs is now easy and quick (at least for defensive access), but a day felt like a lifetime this week. Submit your app now before you really need it.
- Getting Mythos access unlocked a deeper level of analysis that helped us shift from reactive mitigation to proactive patching. Wish we had done this a few weeks earlier. Get started now!
Coming to the service near you!
AI TIP FOR NON-LAWYERS
Health insurance denied a claim? Snap a photo of the denial letter and the coverage section of your policy, upload both to ChatGPT or Claude, and ask:
"Why was this denied, what's my deadline to appeal, and what does my policy say about this treatment? Quote the policy language."
Then have it draft the appeal letter using those quotes.
Two things before you send anything. Black out your member ID and Social Security number before you upload. And check the appeal deadline against the letter yourself, because if the AI gets that date wrong you can lose the appeal entirely.
Worst case, they say no twice.
👏👏We need a dependable repository like this one. Carver can help you with the task you wish to accomplish with this - legal engineering to take this through various approval/analytical/safety/implementation workflows.
Introducing the free Government & Legal Index for ChatGPT & Claude.
Give your AI agents 30M+ US government pages, from federal statutes and court opinions to city codes and permits.
Research tax rules, permits & case law with the Firecrawl plugin.
firecrawl.dev/alexandria/gov…
Dont understand why the NYTimes has to be so narrow minded. He is a well regarded monk with deep knowledge of vedanta, Hindu philosophy. The gray lady should retire.
Swami Sarvapriyananda calls out the New York Times for omitting the Hindu monk from their coverage of the Anthropic AI summit: "Islam, Judaism, Christianity, Buddhism… and Hinduism? New York Times ko yaad hi nahi aaya." #NDTVWorldSummit
"The Indian press picked it up and made it out as if I was the only person there. Well, that’s KARMA for you, New York Times—take that." 🔥
Venkata Pingali retweeted
A legal tsunami is heading for government. They system is not built to handle the volume of input that agents are now able to produce. Without new infrastructure, our institutions will be under permanent DDOS attack. Who wants to build this?
All the news that is fit to print! You cant blame the gray lady for trying to educate the normies with impartiality, fairness, and restraint.
How Rahul Gandhi is transforming from a shy dynast to an effective challenger to Narendra Modi. With @PreetaTweets
nytimes.com/2026/10/18/world…
This extends to pretty much product that hasnt checked the legality of the action in that context.
The litigation exposure is extraordinary for products like Bee. The TOS pushes that liability onto the consumer. Sit next to a billionaire at dinner using a Bee & sue them for every violation at $5,000 a pop. In the EU, it becomes a criminal violation,
triangulatedreality.substack…
Indictment of the 1960-90 socialist policies. Thankfully India came close to bankruptcy in 1990 which led to reforms. This was large scale economic mismanagement.
Coming to your doorstep soon. Expect a flood of requests/business engagement from hyper-rational, persistent, super-intelligent agents.
Large scale violence in capitals across EU. It is happening regularly now. Friends in EU generally pessimistic and discouraging travel.
When building becomes free, moats will move to what remains scarce, contested, or difficult to change. Do you know what your moat is?
This is the fast fashion era of software. AI is rapidly driving the cost of building software toward zero. Code, agents, and sophisticated products are becoming easier to replicate. Product life cycles will shrink, and technology advantages that lasted years may last weeks.
So where will the moats be? Where friction persists:
• Platforms (Ecosystem friction): Network effects, established integrations, and control over access make platforms difficult to displace.
• Distribution (Discovery friction): Customer attention is scarce. Getting discovered, evaluated, and selected remains difficult, even when AI agents become the buyers.
• Scarcity (Resource friction): Capital, compute, physical infrastructure, and exclusive assets cannot be generated simply by writing software.
• Authority (Permission friction): Knowing what to do is different from having the legal or institutional authority to do it.
• Accountability (Liability friction): Someone must bear the consequences of decisions. Contracts, obligations, and liabilities outlive the software.
• Conflicting interests (Incentive friction): Better intelligence doesn't eliminate disagreements between customers, vendors, employees, and regulators.
• Coordination (Change friction): Existing processes, relationships, commitments, and systems create real costs of organizational change.
But AI will attack these moats too. Agents will reduce switching costs, automate coordination, and challenge traditional distribution advantages.
Not all friction creates a moat. Some will disappear. Some will become commoditized. Others will concentrate around powerful platforms.
The next competitive advantage may not be in what you can build, but in what others cannot easily access, agree upon, authorize, or replace.
Have you thought about what will happen when models commoditize your work?
And this doesnt even include the legal issues - not something I would have thought about a year back. Now the liability conversations are ramping up fast.
Well, this is awkward.
Back in March, I questioned the grand idea that AI agents could replace entire engineering teams.
Now Meta is reportedly rebuilding management layers because humans are spending too much time cleaning up after the machines.
Turns out firing the adults and leaving the interns in charge wasn’t much of a strategy…. even when the interns are artificial.
It is workable (large question bank with randomized selection, down to a single individual). Problem is less of the exam. Even in the unleaked exam situation, it doesnt seem to measure skill or address the fact that the most degree holders in India are unemployable. Infosys had to create an entire apprenticeship to deal with poor quality. Now even that is meaningless with AI.
Nandan Nilekani, one of India’s most prominent technology leaders, said the country needs to do away with traditional question papers to combat leaks that have disrupted high-stakes exams bloomberg.com/news/articles/…
Venkata Pingali retweeted
“Kupor described law enforcement coordination as one part of the task force’s broader safety agenda.”
“It will work to audit pledges made through the recent AI accords; conduct a full risk assessment of the technology; and ensure the government has the resources to respond if something goes wrong with a model.”
Kupor: ‘Super Intelligence’ task force will press companies on AI safety politi.co/3W5Ywdb
This is a natural evolution. Human courts and law was not designed to operate at wire speed. All legal institutions have to move 'below the agent' to deal with scale and speed. This is what informed our 'legal infra for agents' framing at Carver.
New essay: To keep human judgment, we need AI courts.
AI-generated filings are starting to overwhelm courts around the world.
The most popular response I’ve heard is to fight AI with AI: If we can just give judges the right tools and training, AI will help them keep pace.
But this strikes me as the worst of both worlds.
Because we want human judges in the loop, we don’t get the speed and scale benefits of AI. Yet because we expect them to resolve a massive number of cases quickly, judges (or their clerks) may end up leaning on AI more than they mean to or should.
This essay offers another way: a parallel AI adjudication track.
An AI track — inspired by AI use in private arbitration but formally part of the court system like magistrate judges — would let courts be explicit about how human judgment is allocated.
Done right, this would allow courts to process large volumes of cases while ensuring human judges devote their full attention to the cases where their judgment is needed most.
Read more in @PaxMachinaMag: paxmachina.ai/to-keep-human-…
