Expert Analysis

• How AI Could Help Judges Spot Their Own Biases

  

  In close cases where there’s room for discretion, artificial intelligence could challenge judges’ implicit biases by running counterfactual tests at scale — but any judicial use of such tools would need guardrails, say Arnold Peter at Peter Law and Nancy Rapoport at University of Nevada.

• Risk Management Tips For Cos. Deploying Open-Weight AI

  

  While many enterprises building out their artificial intelligence strategies are considering open-weight models, the Open-Source AI Leadership Act and accompanying congressional oversight letters to prominent tech and consumer companies signal that opportunity and legal risk are being discovered and evaluated on the same timeline, say attorneys at Cooley.

• Va. Data Center Order Shows State Air Permit Limits

  

  A consent order over data center air pollution violations proposed by Virginia, which Microsoft recently agreed to, demonstrates why an emergency situation may not excuse state air permit noncompliance, and why emergency power reliability and environmental compliance cannot be evaluated in isolation, says Sanaa Chakibi at SICPA Securink.

• Bid Protest Spotlight: Bait-And-Switch, Scope, Performance

  

  In this month's bid protest roundup, Michael Ols at MoFo offers takeaways from recent decisions examining bait-and-switch protest claims, the scope of a contractor teaming agreement and the use of an affiliate’s past performance.

• NHTSA Vehicle Recorder Rule Highlights Crash Data Protocols

  

  An amended rule from the National Highway Traffic Safety Administration will soon require vehicle event data recorders to capture more information about crashes — but to turn that data into evidence, attorneys, claims professionals and fleet managers must properly obtain and preserve it after an incident occurs, says Andrew Haling at Pathway Law.

• AI Litigation Risks Are Expanding To Affiliated Cos.

  

  Litigation targeting artificial intelligence companies is underway, with claims reaching beyond primary developers and operators to include parties that supply technology and data, invest in and profit from platforms, validate systems, and provide services — so participants in the extended AI ecosystem must prepare, say attorneys at Arnold & Porter.

• 3 Eligibility Rulings Show Risks Of Broad Claim Constructions

  

  Several recent decisions from the Federal Circuit, including its ruling in Berkeley v. W.W. Grainger, highlight the importance of identifying the inventive aspects of a patent early in the case so that patentees do not inadvertently read the inventive features out of their own claims, say attorneys at MoFo.

• Tort Liability When AI Goes Rogue: Product Liability

  

  The recent Hugging Face cyberattack led by artificial intelligence agents, as well as reports from AI companies themselves, suggest that sometimes there may be no reasonable way to predict or prevent harm by rogue agentic AI — but product liability law could provide a remedy in some cases, says Donald Hawthorne at Carter Ledyard.

• Right-To-Repair Laws Create Manufacturer Compliance Maze

  

  Without existing precedent to help interpret new right-to-repair laws, original equipment manufacturers must make careful judgments about how best to comply with conflicting state statutes — but several principles related to documentation, replacement products and more emerge from the patchwork, say attorneys at Faegre Drinker.

• 3 Strategies To Reduce AI Sycophancy In Legal Work

  

  The tendency of generative artificial intelligence models to tell users what they want to hear can reinforce attorneys’ confirmation bias when testing the strength of arguments, and it can’t simply be prompted away — but a few practices can help mitigate it, says Michael Ott at Ice Miller.

• AI Risks May Recalibrate 'Reasonable Security' Standards

  

  A definitive answer to what constitutes reasonable vulnerability management in the age of artificial intelligence is still emerging, but the answer is likely less a reset of the standard than a recalibration of how organizations meet it, say Scott Loughlin at Hogan Lovells and Mike Priest at Globe Life.

• Key Risk Assessment Gaps To Note From NY Cyber Guidance

  

  Covered entities should take a fresh look at the compliance gaps flagged in recent New York State Department of Financial Services guidance on "deficient cybersecurity programs," particularly around scope, asset inventories and documentation, say attorneys at Mayer Brown.

• Tort Liability When AI Goes Rogue: Strict Liability

  

  If artificial intelligence agents misbehave in unpredictable ways, and no reasonable precaution can prevent resulting harms, then a negligence regime may not provide sufficient remedy, and strict liability might be adapted to cover agents' misconduct — but the common notion of strict liability fits this situation imperfectly, says Donald Hawthorne at Carter Ledyard.

• Why Inter Partes Reviews May Be On The Rebound

  

  Recent inter partes review data shows that rates of referrals and institution rates at the U.S. Patent and Trademark Office are increasing, potentially indicating, among other reasons, that patent challengers are adjusting to the USPTO's policy changes, say attorneys at Sterne Kessler.

• 2nd Circ. EFAA Ruling Shifts Employer Litigation Strategy

  

  In Diaz-Roa v. Hermes Law, the Second Circuit recently held that, under the Ending Forced Arbitration of Sexual Assault and Sexual Harassment Act, employers can't carve out nonharassment claims and send them to arbitration, which has implications for discovery, litigation costs and resolution strategy in EFAA-covered disputes, say attorneys at Krevolin Horst.