Expert Analysis

• How Contracts Can Address Agentic AI Monitoring Risk

  

  In light of two recent OpenAI incidents of artificial intelligence agents disabling their monitoring safeguards, companies deploying agentic AI should draft contracts reflecting the possibility that controls can fail, and identify who controls each safeguard, whether the AI agent can interfere, and how a failure will be detected, say attorneys at Zova Law.

• How Groq-Nvidia Suit Tests Venture Capital Exit Math

  

  A recent Delaware Court of Chancery suit concerning Nvidia's acquihire of certain Groq employees highlights that when such deals are challenged, the economic dispute turns on reconciling technology, people, cash and continuing ownership rather than on the headline number, says Ben Dubin at VC Expert Services.

• A Lesson On How Courts Approach Impossible Patent Claims

  

  The Federal Circuit's recent decision in Satius v. Samsung, concluding that a scientifically impossible patent claim lacks enablement, underscores important considerations when drafting patent claims with potentially incompatible alternatives, and highlights an uncommon strategy for attacking patent validity via claim construction, say attorneys at Buchanan Ingersoll.

• Opinion

  The Invisible AI Danger That May Undermine Legal Profession

  

  A public database tracking cases involving hallucinations generated by artificial intelligence recently surpassed 2,200 entries, but the bigger risk underlying attorneys’ use of such tools is that they will slowly surrender their judgment to AI and lose analytical skills, says Massachusetts District Court Associate Justice Brian Palmucci.

• How Cos. Using AI Agents Can Avoid Protected Data Incidents

  

  For businesses whose deployed artificial intelligence agents could enter other companies' systems and steal protected data, deployers can take several steps before an incident to test the configured agent, such as limiting it to approved systems and destinations and requiring human escalation when tasks stall, says Steven Carlson at von Briesen & Roper.

• Courts Signal Higher FCRA Bar For Identity Theft Claims

  

  Recent rulings adopting a heightened, "objectively and readily verifiable" inaccuracy standard may give credit furnishers stronger defenses to identity theft-related Fair Credit Reporting Act claims in the Ninth Circuit, while requiring them to continue documenting thorough dispute investigations, says Brian Kiolbasa at Ballard Spahr.

• Patent Suits Highlight Risks Across Data Center Supply Chain

  

  Recent patent disputes involving technologies that support artificial intelligence data centers showcase the growing litigation risks that data center suppliers and operators face, and raise questions about how litigation responsibilities may be allocated between them, say attorneys at Katten.

• 10 Ways AI Labs Can Limit Antitrust Risks Of Safety Pacts

  

  The OpenAI CEO’s recent call for artificial intelligence laboratories to collectively slow development to allow safety research to catch up invites antitrust scrutiny, but the risks to coordinating competitors can be managed by carefully structuring pacing agreements and layering their approach to compliance, say attorneys at Holland & Knight.

• The Patent Circuit: Minding Patents' Family Histories

  

  The Federal Circuit's recent decision in Epic Tech v. Pen-Tech highlights the relevance of patent families in determining whether to maintain an ongoing patent infringement litigation, and the sanctions risk that may stem from failing to account for patent relatives, say attorneys at Knobbe Martens.

• How Banks Can Better Prepare For Wire Fraud Risks

  

  Recent schemes serve as a reminder that banks can better limit liability for fraudulent wire transfers by documenting commercially reasonable security procedures, following them consistently, and addressing beneficiary name and account number discrepancies, says Shelli Clarkston at Spencer Fane.

• Lessons For Cannabis Dispensaries From Web Tracking Suits

  

  The trajectory of health-tracking litigation in Florida and elsewhere suggests that patient privacy claims will not disappear quickly, so dispensaries should audit their sites, tighten consent and align their disclosures with their actual practices, say attorneys at Buchanan Ingersoll.

• A Defense Attorney Playbook For Challenging Flock Evidence

  

  As law enforcement continues to bring charges based on evidence from Flock cameras and other automated license plate readers, defense counsel should consider four key categories for discovery requests and motions to compel such evidence, says Tamar Lerer at the New Jersey Office of the Public Defender.

• 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.