Patrick Lonergan

CIPA Litigation Lacks Clarity In California The California Invasion of Privacy Act (CIPA) is being used to challenge web tracking, and the results are mixed: ✅Preliminary Judgements in State and Federal Courts Some state courts have dismissed CIPA claims finding that pixels and cookies don’t meet the definition of “pen registers.” Some federal courts have allowed CIPA claims to proceed where they involve more invasive tech like session replay which records a website visititor’s screen while they interact with a website. ✅Recent Cases In November the Adidas case survived summary judgment in federal court because allegations of Adidas’ use of tracking pixels to collect website visitors’ personal information was strong enough to proceed under CIPA. The Ink America Int’l case was dismissed in state court in December. The decision cited, in part the legislative intent of CIPA to address surveillance and not to broadly regulate the normal operation of commercial websites. ✅Possible Clarification Through Legislation SB 690 proposes clarifying CIPA by exempting commercial business purposes, borrowing language from the CCPA. While widely supported, it stalled in Assembly last year and has been converted to a two-year bill. Perhaps we will see the bill pushed forward in 2026, but uncertainty remains. 🗝Takeaway CIPA litigation isn’t going anywhere anytime soon, and there is no business-size threshold for applicability like with CCPA. Consider your CIPA obligations, particularly where session replay, voice/video transcription, and sensitive data collection are involved. As it currently stands, CIPA relies on an “opt in” approach to tracking, so consider “opt-in” rather than CCPA “opt-out” style consent. 2025 California Privacy Review: https://bit.ly/4b62UhZ Adidas Class Action:https://lnkd.in/gXQs9QYD Photo by the Author

18

3 Comments

𝐃𝐚𝐲 𝟑 — AI Hallucinations in Law Unit + integration tests validate logic paths—not whether the document’s structure lies. 🧪 𝐖𝐡𝐚𝐭 𝐭𝐞𝐬𝐭𝐬 𝐜𝐡𝐞𝐜𝐤: • Clean, aligned sample contracts (well-formed fixtures) • Same input gives the same result every time (deterministic input→output) • Model metrics on a cleaned, hand-picked document set (curated corpora) • Checks that assume unique headings and stable cross-references (schema validators) 🚫 𝐖𝐡𝐚𝐭 𝐭𝐡𝐞𝐲 𝐦𝐢𝐬𝐬: • Headings that contradict the content (mislabels, merges, renumbering) • Clause drift from messy redlines and definitions left without anchors (orphaned definitions) • Copy/convert glitches like reflow, hyphen joins, and lost bullets (OCR/PDF→Word artifacts) • Numbering/style changes (e.g., 1.2 → 1(b)) and broken internal links (cross-reference rot) 🧯 𝐐𝐀 𝐚𝐧𝐭𝐢-𝐩𝐚𝐭𝐭𝐞𝐫𝐧𝐬 𝐭𝐨 𝐰𝐚𝐭𝐜𝐡: • Overfitting to one “perfect” reference contract (“golden doc”) • Treating headings as the label instead of reading the clause itself • Document-level pass rates that hide clause-by-clause misses (clause-level recall gaps) 🧩 𝐒𝐲𝐦𝐩𝐭𝐨𝐦𝐬 𝐢𝐧 𝐩𝐫𝐨𝐝: • Limitation of Liability, Indemnification, or Termination mis-scoped or skipped • High overall score but blind spots on high-impact clauses • Escalations cluster under “Miscellaneous” and “General Terms” 🛠️ 𝐀 𝐨𝐧𝐞-𝐰𝐞𝐞𝐤 𝐮𝐩𝐠𝐫𝐚𝐝𝐞 𝐭𝐨 𝐲𝐨𝐮𝐫 𝐭𝐞𝐬𝐭 𝐬𝐭𝐚𝐜𝐤: • Create intentionally scrambled versions of the same contract (structural adversarial set)—keep substance constant • Use two readers: one follows formatting, one follows meaning (dual-path parsing); compare results when they disagree (arbitrate) • Add detectors: flag heading↔content disagreements (divergence) and broken references/definitions (consistency checks) • Turn on logs that show which cues the AI followed (observability); alert on Limitation of Liability/Indemnification/Termination anomalies • Track accuracy drop from clean→scrambled and keep it under 10% (Δ-accuracy); measure how often each clause is correctly caught (clause-level recall); drive the disagreement rate down (divergence rate ↓) ✅ 𝐑𝐞𝐚𝐝𝐢𝐧𝐞𝐬𝐬 𝐠𝐚𝐭𝐞 𝐛𝐞𝐟𝐨𝐫𝐞 𝐠𝐨-𝐥𝐢𝐯𝐞: • Model follows substance over structure under intentional structure changes (perturbations) • On conflict, raise a warning instead of guessing (uncertainty flag) and route to review 𝐁𝐨𝐭𝐭𝐨𝐦 𝐥𝐢𝐧𝐞: If QA never attacks structure, your AI can “pass” and still fail on Tuesday’s messy contract. 𝐓𝐨𝐦𝐨𝐫𝐫𝐨𝐰: How to adversarial-test your legal AI (SCRAMBLER in practice). #LegalTech #AI #ContractReview #InHouseCounsel #Procurement

4

The potential of AI in the Legal space is huge, but the story is a far more complicated. This is such a great article describing the complexity myself and other in house Legal leaders are facing in trying to bring the value of AI to our teams.

22

2 Comments

For folks using Generative AI, be wary of seemingly accurate yet incomplete responses. While AI tools seem to excel at often producing technically correct information, they are also limited by their dataset and often miss critical context and nuance. Consider legal research tools that correctly cite laws but overlook recent precedents that completely change the case outcome. This isn't about AI being wrong; it's about AI being right in ways that aren't quite right enough. This incompleteness challenge stems from fundamental limitations: training data that inevitably becomes outdated, algorithms that struggle to distinguish essential from ancillary information, and the black box problem where we can't see what's been omitted. Rather than viewing this as a roadblock to AI adoption, smart professionals are recognizing it as an opportunity to develop crucial verification and supplementation skills. We should be less focused on AI replacing human judgment and more mastering combining AI with context and judgment. I’m Colin, General Counsel of Malbek - CLM for the Enterprise and author of The Legal Tech Ecosystem. #legaltech #innovation #law #business #learning

33

17 Comments

The Supreme Court just punted on one of the biggest IP questions of the AI era. And honestly? I’m not mad about it. On March 2, SCOTUS denied cert in Thaler v. Perlmutter — the case where computer scientist Stephen Thaler tried to register copyright in an image created entirely by his “Creativity Machine” AI, with zero human involvement. The DC Circuit said no. The Copyright Office said no. And now the Supreme Court has said: “We’re not ready to talk about this yet.” Here’s what practitioners need to understand: The human authorship requirement isn’t going anywhere. If a human didn’t exercise creative control over the work, it doesn’t get registered. Period. But here’s the nuance most people are missing: This case was about purely autonomous AI output. No prompting. No curation. No editing. Thaler literally listed the AI as the author on the application. That’s not how most of us are using AI. If you’re prompting, directing, selecting, editing, you’re still the author. The Copyright Office has registered hundreds of AI-assisted works where a human was in the creative loop. The door isn’t closed. It’s just not open for machines flying solo. For my fellow attorneys: this is a documentation problem now. If your clients are using AI in creative workflows, they need to be logging prompts, tracking editorial decisions, and documenting human involvement at every stage. The registration battle will be won or lost in the paper trail. For the AI builders — this isn’t over. The Court declined cert. They didn’t rule on the merits. A better vehicle will come along. Probably one involving AI-assisted work where the line between human and machine creativity gets genuinely blurry. That’s when it gets interesting. The law is playing catch-up. As usual. #AI #Copyright #LegalTech #SCOTUS #IntellectualProperty #AILaw

5

1 Comment