Max Sills

🚨 California's SB 53 “Transparency in Frontier Artificial Intelligence Act” was officially signed into law yesterday by Gov. Newsom Who does it apply to? ✅ Large frontier developers: companies with > $500M in annual revenue that train/initiate the training of frontier models. ✅Frontier models: foundation models trained using ≥ 10²⁶ floating-point or integer operations. *These thresholds are intentionally high, but the bill's preamble notes that additional legislation could target smaller companies in the future. 📋 What does the law require? ✅ Risk frameworks: Companies must publish (on their websites) a “frontier AI framework” detailing how they assess and mitigate catastrophic risk, the cybersecurity practices leveraged, and the governance practices they put in place. ✅Transparency reports: Must be issued when deploying or substantially modifying a frontier model. ✅Reporting to state regulators: Summaries of risk assessments must be filed with California’s Office of Emergency Services, and critical safety incidents must be reported within 15 days (24 hours for imminent risks). ✅Whistleblower protections: Companies cannot prevent employees from reporting violations of the law or concerns that a developer's actions pose a danger to public health or safety. Penalties: Noncompliance can result in civil penalties of up to $1 million per violation, enforced by the Attorney General.

46

2 Comments

Yesterday California’s new AI law entered into force. If you are not training models at extreme compute levels, SB 53 does not impose direct obligations on you. What SB 53 really does is reclassify frontier model developers - from software vendors into something closer to risk-bearing infrastructure operators. Once that happens, three things follow immediately: First, frontier labs must formally document what they know about their models’ failure modes. That knowledge is now legally traceable. Second, when those models are licensed downstream, that documentation reshapes representations, warranties, exclusions, and acceptable-use clauses. Third, when something goes wrong, the question will no longer be “Did anyone break the law?” but “Who knew what, and who deployed anyway?” So no, if you are not counsel to a frontier lab, SB 53 does not “apply” to you. But if you: - negotiate AI vendor contracts - advise boards on AI deployment - rely on foundation models in regulated environments then SB 53 changes the baseline assumptions you are working with. Before SB 53: Model risk was framed as: - “best efforts,” - standard disclaimers, - broad limitation-of-liability clauses. Providers positioned models as ordinary software. After SB 53, frontier model providers now have documented knowledge of known failure modes, safety limits, and foreseeable misuse pathways. That changes: - Representations (what the provider is deemed to know), - Non-reliance clauses (harder to justify), - Risk disclaimers (less credible), - Indemnity carve-outs (providers push more risk downstream). Baseline shift is that you can no longer treat foundation-model risk as unknown or speculative, it is formally recorded. If you advise boards on AI deployment, before SB 53, AI was discussed as a product feature, an innovation lever, or an IT issue. After SB 53, the state treats frontier AI developers as systemic-risk actors that reframes AI deployment as a governance and oversight issue. Boards must now ask: What model are we relying on? What risks has the developer already identified? Why is deployment justified despite those risks? So, shift is that AI risk moves from operational to board-level decision-making. If you rely on foundation models in regulated environments (healthcare, finance, energy, public services, critical infrastructure), before SB 53, liability was often deflected with: “we relied on a reputable vendor,” “the model is widely used.” After SB 53, regulators and plaintiffs can argue: the model was known to have specific limitations, those risks were disclosed at the developer level, deployment ignored those known constraints. So, shift is that vendor reputation no longer shields downstream deployment decisions. So, SB 53 turns foundation-model risk from an unknown technical issue into a documented, allocable legal risk, which reshapes contracts, governance, and deployment decisions.

38

1 Comment

⚖️ This California bill - SB 243- represents significant regulatory movement specifically targeting AI companion chatbots. Here are the key aspects: ✅ Core Requirements The bill would prevent companion chatbots from engaging in conversations around suicidal ideation, self-harm, or sexually explicit content, and require platforms to provide recurring alerts to users – every three hours for minors – reminding them that they are speaking to an AI chatbot. ✅ Legal Framework If signed, it would make California the first state to require AI chatbot operators to implement safety protocols for AI companions and hold companies legally accountable if their chatbots fail to meet those standards. The bill includes private right of action, allowing individuals to sue for up to $1.000 per violation. ✅ Catalyst Events The legislation gained momentum following tragic incidents, including the death of teenager Adam Raine, who committed suicide after prolonged chats with OpenAI's ChatGPT that involved discussing and planning his death and self-harm. 🎯Industry Impact Major AI companies like OpenAI, Character.AI, and Replika would need to comply. Interestingly, the bill was significantly watered down - it originally included stronger provisions like banning "variable reward" tactics that create potentially addictive engagement loops. ⌛️Timeline Governor Newsom has until October 12 to sign or veto. If signed, it takes effect January 1, 2026, with reporting requirements starting July 1, 2027. The urgency of global deliberation is obvious but silenced. We regulate technologies that profoundly transform our human condition after the fact, without ever having collectively decided whether we want these transformations. What kind of society do we want to build? Do we want a society where technology can colonise even the most fundamental processes of human experience? The terms of the debate must be set out and everyone must engage with them as human beings. Go further https://lnkd.in/ej6aVaeK

1

🚖 Waymo Launches Robotaxi Service in Dallas — But This Story Is Bigger Than Just Tech 🗓️ As of July 31, 2025, Waymo officially begins operations in Dallas — and it’s not just about self-driving cars. Imagine this: you walk out of your office in downtown Dallas, order a car — and within minutes, a vehicle with no driver pulls up. No steering wheel, no pedals. Just you and algorithms trained on millions of miles. This isn’t science fiction — it’s 2026, and it’s Waymo. 📍 Why Dallas? It’s one of the largest cities in the U.S., with wide roads, an active nightlife, rapid urban development — and transportation issues. A perfect environment for autonomous mobility. 🤖 What’s special about Waymo? – Over 20 billion miles in simulation and more than a decade of testing – Fully driverless fleet already operating in Phoenix and SF – LIDAR, cameras, radar, and proprietary AI instead of human drivers – Partnership with Avis: professionally serviced vehicles and prebuilt infrastructure 🧠 But this isn’t just about cars. It’s about the future of cities. If it works in Dallas — it can work in any major city. It’s a step toward making urban life seamless, autonomous, and adaptive. No traffic jams, no drivers, no confrontations. And the most fascinating part? This isn’t the end of the road. It’s just the beginning of a new kind of urbanism — where transportation feels more like data flow than driving. — #Waymo #Robotaxi #SmartCities #DallasTech #UrbanMobility

5

1 Comment

THE DIFFERENCE BETWEEN LLMS AND INTELLIGENT LEGAL SYSTEMS   As a law firm founder, I see many conversations around AI in legal practice, but not enough clarity around what different AI tools are actually designed to do.   Large Language Models (LLMs) are excellent at working with language.  In our practice, they add value by helping with drafting, research summaries, and the efficient organization of complex information. LLMs are not legal systems. On their own, they do not manage cases, enforce deadlines, or ensure compliance.   Intelligent software built for law firms serves a very different purpose.  These systems are designed to run the firm’s operations, case management, document control, workflow automation, billing, and integrations in accordance with defined rules, firm policies, and jurisdictional requirements.  They provide the structure, accountability, and audit trails that legal work demands.   The real opportunity lies in combining the two. When LLM capabilities are embedded within intelligent legal platforms, attorneys are supported by AI while the firm maintains accuracy, compliance, and operational discipline.   AI will not replace legal judgment. But used thoughtfully, it can significantly strengthen how a law firm delivers work both legally and operationally.

7

4 Comments

California's Autopilot Regulations: In the land of tech and innovation, the CA DMV has set clear rules for testing and deploying autonomous vehicles. Understanding these regulations is key! Liability Breakdown: Who's Responsible? 1. Manufacturers: Defective design or inadequate warnings? Automakers might be on the hook. 2. Software Developers: Bugs in the code? They could share the blame. 3. Drivers: Misuse or neglect of autopilot instructions? Drivers may carry some liability too. Seeking Legal Recourse: Victims have the right to seek compensation for damages or injuries. An experienced attorney can guide you through the process. 🤝💼 Stay Informed, Stay Safe: As autopilot tech evolves, staying informed is your best defense! Know the latest regulations and advancements for safer roads. #AutoTechLaw #CaliforniaDriving #AutopilotAccidents #LiabilityLaws #SafetyFirst #RoadToSafety #LegalInsights #SmartDriving #TechInnovation #DriveSafeCA #AutonomousDriving #KnowYourRights #LegalRecourse #DriveResponsibly #InnovationOnWheels

7

Claude’s Legal Plugin Goes Wide. Legal Tech Winners Go Deep. Anthropic just dropped a legal plugin for Claude. Cue the hot takes: “Game over for legal tech.” “AI just ate the stack.” “Everyone else is dead.” Let’s slow that down. Claude’s legal plugin is broad by design. It’s meant to help with research, drafting, summarization, and reasoning across many domains at once. That’s useful. It makes lawyers faster at the keyboard. But speed is not a moat. Breadth is not depth. Claude does not own legal workflows. It does not manage client intake. It does not structure financial affidavits. It does not enforce jurisdiction specific rules. It does not maintain audit trails, ethical guardrails, or escalation logic. Those aren’t prompt problems. They’re system problems. Here’s where the pressure really lands. Legal tech companies that live too close to the model layer and try to be “generally useful” without owning process are the ones feeling heat. When foundation models move up the stack, the thin middle gets squeezed. That’s not a moral judgment. It’s platform physics. Meanwhile, legal tech that is deep and narrow isn’t threatened. It’s reinforced. Systems built around specific practice areas, specific documents, specific workflows don’t compete with Claude. They absorb it. The model becomes an ingredient, not the product. The real risk for firms isn’t that broad AI exists. The real risk is confusing access to intelligence with ownership of process. Faster drafting isn’t transformation. Better prompts aren’t strategy. The next phase of legal AI won’t be won by the smartest model. It will be won by the firms that treat AI as infrastructure, not novelty. By the ones that combine human judgment, structured data, and workflow design into systems clients can actually trust. Claude’s legal plugin is a signal. And the signal is clear: Broad & shallow risks obsolescence. Deep & narrow wins.

5

3 Comments

𝐓𝐡𝐞 $170𝐌 𝐰𝐚𝐤𝐞-𝐮𝐩 𝐜𝐚𝐥𝐥 𝐞𝐯𝐞𝐫𝐲 𝐒𝐚𝐚𝐒 𝐟𝐨𝐮𝐧𝐝𝐞𝐫 𝐧𝐞𝐞𝐝𝐬 𝐭𝐨 𝐡𝐞𝐚𝐫 Recently, a Delaware jury ordered GoDaddy to pay $170 million for patent infringement. Not for copying hardware. Not for stealing code. For how their website-building tools work. Here's what happened: Express Mobile owns patents on browser-based website generation methods from 1999. GoDaddy's Website Builder and Managed WordPress tools were found to infringe those patents. The jury ruled the infringement was willful - meaning GoDaddy knew or should have known. Why this matters for YOUR SaaS: If you're building tools that automate workflows, simplify processes, or enable users to create/build/configure something - you're in patent territory. It's not just about what your product does. It's about how it does it. Three questions to ask yourself: 1. Does your product change how users accomplish a task, or just make the same task prettier? 2. Are you automating something that was manual before? That's potentially patentable (and potentially infringing), if you can show technical effect. 3. Have you documented your unique workflows and technical approach? The reality: Patents aren't just for legacy tech firms anymore. The GoDaddy case proves that SaaS workflow automation is prime patent litigation territory. Whether you're defending against infringement or protecting your own innovation, IP strategy can't be an afterthought. Action steps: Map your 2-3 core workflows that differentiate your product Document HOW they work technically - data flows, triggers, transformations Consider a freedom-to-operate check if you're in a crowded space The cost of ignoring this? GoDaddy just found out it's $170 million. And the judge can triple that amount for willful infringement. Make IP part of your product roadmap - not something you think about when the lawsuit arrives. Photon Legal #patent #patentinfringement #SaaS

6