Our latest technical research paper: Neural patent search systems typically require examiner citation data for training—expensive and potentially biased. Our work implements a self-supervised alternative: training BERT to match patent claims with their own description passages. Tested on 6,447 USPTO patents, our approach achieves 96% accuracy on claim-description matching without any examiner labels, demonstrating that document structure alone can teach models to recognize patent-relevant content. cipher.law/s/CSML_Paper_Fina…

AI Patent Eligibility Dramatically Expanded at USPTO cipher.law/research/ai-paten…

The USPTO's latest memo fundamentally shifts AI patent examination: examiners must now prove greater than 50% probability of ineligibility before rejecting claims. No more rejections based on uncertainty. The "close call" standard means tie goes to the applicant - a significant practical change for AI patents. This guidance, combined with new limits on 'mental process' rejections that had threatened hardware-based AI operations, creates the most favorable AI examination environment since before the July 2024 confusion that destabilized neural network training claims. cipher.law/research/navigati…

We're drafting patents in 3 days that used to take 3 weeks. And covering 5x more embodiments than legacy drafting techniques. This is the new reality of patent preparation when you combine expertise with the right AI workflows. linkedin.com/feed/update/urn…

Amendments to Executive Orders 14144 and 13694 introduce deadlines for post-quantum cryptography adoption, requiring agencies to implement TLS 1.3 or equivalents by 2030 and publish compliant product lists by year's end. For AI, the order mandates treating systems as critical assets, addressing risks like prompt injection in vulnerability management—aligning with broader efforts to secure software supply chains via updated SP 800-218 guidelines. These changes could reshape patent strategies in quantum tech and AI security, urging innovators to prioritize compliance in filings.

USPTO grants another 6 AI-related patents to @IBM today

TO WIN THE AI RACE, THE AI DIFFUSION RULE MUST GO The Trump administration has announced its intention to repeal the Biden administration’s AI Diffusion Rule. As the Commerce Department’s Bureau of Industry and Security (BIS) confirmed in a statement yesterday: “The Biden AI rule is overly complex, overly bureaucratic, and would stymie American innovation. We will be replacing it with a much simpler rule that unleashes American innovation and ensures American AI dominance.” This is an excellent decision by Secretary of Commerce @howardlutnick and Under Secretary of BIS Jeff Kessler. There were several major problems with the Biden Diffusion Rule: 1. Overreach of Export Control Authority First, the rule marked an unprecedented—and arguably unlawful—expansion of export control authority. Under the Export Control Reform Act (ECRA) of 2018, the President is empowered to restrict exports of dual-use technologies that have both civilian and military applications. That authority has been used to restrict the sale of advanced semiconductors to China, a policy with broad bipartisan support. But the Diffusion Rule went significantly further. It required nearly all global sales of high-end GPUs—even to trusted allies—to obtain export licenses or fit into a narrow set of license exemptions. This forced much of the global data center and AI infrastructure industry to seek approval from Washington, creating a bottleneck that chilled legitimate, non-sensitive commerce. 2. Bureaucratic Allocation of Compute The rule imposed detailed numerical caps on how many chips and how much computing power foreign entities could acquire and operate. This was a radical departure from market-based allocation principles, placing the U.S. government in the position of rationing compute power globally. It effectively turned Washington into a central planner for the global AI industry. 3. Alienation of U.S. Allies The rule also strained relationships with key allies by arbitrarily dividing countries into compliance “tiers,” labeling many friendly nations as second-class partners. This kind of regulatory hierarchy undermines trust and risks pushing allies toward non-American technology alternatives. 4. Lack of Due Process The Diffusion Rule was issued just five days before the end of the Biden administration without a meaningful public comment or review period. Given its sweeping scope, its retroactive elements, and the global compliance burden it imposed, this rollout was deeply flawed from both a procedural and practical standpoint. ⸻ In his first week in office, President Trump directed us to win the AI race. The Biden Diffusion Rule undermines that goal. It bogs American tech companies down in red tape, while slowing the global adoption of U.S. technologies at a time when we should be encouraging the world to build on our tech stack. As @VP J.D. Vance emphasized in his Paris Speech on AI, the United States should be the gold standard and partner of choice for our allies and strategic partners. If we make it too hard for them to work with us, we risk pushing them into China’s orbit. China has already launched a Digital Silk Road as part of its broader Belt and Road Initiative. If we don’t offer a compelling alternative, we leave the field open. Yes, we must take aggressive steps to prevent advanced semiconductors from being illegally diverted into China. But that goal should not preclude legitimate sales to the rest of the world as long as partners comply with reasonable security conditions. Today, American chips remain superior to China’s—but that lead is narrowing. If U.S. companies are hamstrung by excessive regulation, and foreign customers are blocked from buying our technology, we risk ceding global markets and influence to Chinese competitors. Right now, we have the opportunity to entrench the American tech stack worldwide while we still have a commanding lead. Let’s seize it.

USPTO awards to @IBM another 7 AI-related patents today for wide range of inventions. IBM might not be in the headlines with the more consumer-facing companies, but is building a massive AI patent portfolio.

AI Patent Law Update: the Federal Circuit's decision in Recentive makes it clear that simply waving the "AI-wand" and sprinkling "AI-dust" over patent applications will not satisfy USPTO subject matter requirements. Big premium now on being able to explain the technical improvements - in detail - in the application. cipher.law/research/navigati…

Navigating the AI Patent Landscape: A 2025 Update on USPTO Guidance and Best Practices. cipher.law/research/navigati…

Tech Law Update: Navigating New State Laws for Terms of Service. Recent Terms of Service (ToS) developments and their implications for business. ToS is not just a legal shield anymore - it's a compliance artifact. cipher.law/tos-update

Well done @SolveIntel on the $12M raise! Been on the platform since late 2023 and it has become an absolutely mission-critical part of our patent counseling practice. Better patents - faster. Looking forward to what you are able to build next! solveintelligence.com/blog/p…

OpenAI calls DeepSeek ‘state-controlled,’ calls for bans on ‘PRC-produced’ models tcrn.ch/3DGJ1ji

New NVIDIA Patent Analysis Dashboard released: cipher.law/nvidia-patent-das…

US Patent Office awards @Tesla broad patent on predicting distance, direction, and velocity of other vehicles and responsively controlling the vehicle in autonomous driving. patentcenter.uspto.gov/appli…

Google awarded 12 new AI-related patents today covering: • Content quality prediction • Speaker audio separation • Neural network speech generation • Semantic object creation • Image extension • Voice interaction services

Updated:: Federal Law and Policy Tracker for Artificial Intelligence docs.google.com/document/u/2…

What a smart statistics department would do right now is rename itself the AI department and rake in the students and dollars.