Writing ยท AI / Automation / Tech

2026-01-16
๐€๐ˆ ๐‰๐ฎ๐ฌ๐ญ ๐๐ซ๐จ๐ค๐ž ๐ญ๐ก๐ž ๐‹๐š๐ฐ (๐‡๐ž๐ซ๐žโ€™๐ฌ ๐–๐ก๐š๐ญ ๐๐ซ๐ž๐š๐ค๐ฌ ๐๐ž๐ฑ๐ญ) Brian Wangโ€™s article covers five AI legal issues: IP ownership, data privacy, algorithmic bias, liability for errors, and workforce displacement. Wang covers real problems. Here are some additional issues with AI & Law. ๐“๐ก๐ž ๐‹๐ข๐š๐›๐ข๐ฅ๐ข๐ญ๐ฒ ๐‚๐ก๐š๐ข๐ง ๐ƒ๐ข๐ฌ๐š๐ฉ๐ฉ๐ž๐š๐ซ๐ฌ Your lending AI uses credit scores from another companyโ€™s AI. That AI pulled data processed by a third AI. Someone gets denied unfairly. Whoโ€™s responsible? Legal systems assume clean causation lines. AI creates webs. Courts will demand answers that donโ€™t exist. ๐ƒ๐ข๐ฌ๐œ๐จ๐ฏ๐ž๐ซ๐ฒ ๐๐ž๐œ๐จ๐ฆ๐ž๐ฌ ๐ˆ๐ฆ๐ฉ๐จ๐ฌ๐ฌ๐ข๐›๐ฅ๐ž Opposing counsel demands to see how your AI decided. But the modelโ€™s training data is proprietary. The weights are unexplainable even to creators. Courts will order disclosure anyway. Companies will claim trade secrets. Standoff. ๐‚๐จ๐ง๐ญ๐ซ๐š๐œ๐ญ ๐‹๐š๐ฐ ๐†๐ž๐ญ๐ฌ ๐–๐ž๐ข๐ซ๐ If your AI agent negotiates and executes contracts, when are you bound? What if it agrees to terms youโ€™d never accept? โ€œI didnโ€™t authorize that.โ€ meets โ€œyour AI had apparent authority.โ€ Courts wonโ€™t have an answer. ๐–๐ก๐š๐ญ ๐๐ซ๐ž๐š๐ค๐ฌ ๐๐ž๐ฑ๐ญ Insurance policies werenโ€™t written for AI risk. Exclusions donโ€™t contemplate algorithmic failures. Insurers will exit or charge premiums that lock out small players. That means big tech. Only the largest companies can afford AI compliance infrastructure. Theyโ€™ll help write the regulations. Small competitors get locked out. Innovation slows. The existing players love it. Employment law becomes a minefield, then disappears. Short term: litigation explosion. Medium term: courts require human review of AI decisions. Long term: companies make the human reviewer a $15/hour rubber stamp. Criminal law has no answer for AI fraud. Prosecutors need someone to charge. Theyโ€™ll pick whoeverโ€™s convenient. Low-level employees become scapegoats. Weโ€™re not adapting law to AI. Weโ€™re discovering that modern law assumed humans were in the loop. Remove that assumption and the whole system wobbles. Courts will create answers. Theyโ€™ll be messy, inconsistent, and expensive. The companies with the deepest legal teams win. No surprise there... Everyone else picks through the wreckage.โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹ https://lnkd.in/edq3J7is
AI / Automation / TechOperations / Property ManagementHiring / People / LeadershipMindset / Mental Models / Decision MakingSales / Negotiation

View original on LinkedIn

โ† Back to writing