Mark ReCupido

Update on my current assessment of "Legal AI." Things it does reasonably well: document review and summarizing, drafting assistance, and applying black letter law to relatively straightforward factual scenarios. Things it is still incredibly bad at: Literally everything else. Improvements in the past year: It is even better at the things that it does reasonably well. Improvements still needed: Not confidently being extremely, horribly, laughably wrong when asked to do anything that it doesn't do very well. The biggest delusion currently making the rounds in the legal world: AI is on the verge of (or even currently capable of) replacing young associates. At best, Legal AI is equal to the worst starting associate you can possibly imagine. Except that you might be able to get the associate to improve. You can't make AI do better. (Maybe someone can, but YOU can't.)

52

14 Comments

Hot take on getting people interested in privacy at your organization = GOSSIP! Okay... not gossip in the general sense of the word, but juicy stories. You kept reading didn't you? In my first privacy job, I was the privacy consultant for 35 organizations. They all paid a little bit of my salary for the opportunity to have their local person use me as a resource and do some behind-the-scenes work. Since they paid me, I had to show value. Each quarter their CFOs and CIOs would get together for a meeting and I would be on the agenda. The first few times I attended this meeting, I presented the stats (X number of audits, X number of breaches, etc.). BORING. I had little impact at that meeting. No one remembered me. I’m not sure how much value I was showing despite the work I was doing. I dreaded it. I decided I needed to do something different. I started combing through the case law and prepping before each meeting. I’d find the juicy stories and relate them back to things they needed to put in place at their organizations to ensure compliance. I’d describe the consequences for rogue employees, what happens when your policies are inadequate, the fines (CFOs especially perked up for this one), training requirements, etc. All from the stories of real cases. I made it interesting AND relatable. These C-Suite leaders started talking to me. I even had many people tell me they were always most excited for my speaking time in the meeting. I actually started to look forward to these meetings, which for a highly introverted person is saying something! We can make privacy interesting and impactful. No one wants to hear the legalese. No one gets the privacy jargon. No one cares how many audits you did except for the regulator. People get stories. If you want any of your privacy topics to stick, weave them into a juicy, gossipy, story. And I wouldn’t be a good privacy person if I didn’t mention that you always need to be mindful of privacy when sharing! #privacy

27

5 Comments

Earlier today, the Virginia Senate passed the High-Risk AI Developer & Deployer Act (HB 2094), following its passage by the House earlier this month. If signed into law, Virginia would become the second state to enact comprehensive AI legislation, less than a year after the enactment of the Colorado AI Act. Despite certain differences, the two Acts adopt similar risk-based frameworks: both impose disclosure and notice requirements, provide rights for consumers affected by AI-based decisions, and establish developer and deployer duties of reasonable care to prevent algorithmic discrimination. More broadly, HB 2094's passage and potential enactment signal a growing U.S. consensus around Colorado-style AI regulation, with over a dozen other states considering similar bills this year. In the absence of federal action, states are the prime movers of the emerging U.S. AI regulatory regime. #AIlaw #AIpolicy #AIlegislation #VirginiaAILaw #ColoradoAIAct

31

4 Comments