Snippets.

// AI Briefing · October 17, 2026

AI Briefing

Watch on YouTube ↗Subscribe ↗

Meta's updated commercial terms, which take effect October 30, let it use crawlers to collect content from the websites of businesses with Meta accounts, and they explicitly cover "AI at Meta." Legal experts told Nieman Lab the language could override robots.txt and get Meta training data worth millions in licensing fees for free. A Goldman Sachs prime brokerage note shows global hedge funds suffering steep losses as AI trades unwind violently, with Asia-focused stock pickers hit hardest after South Korea's Kospi fell 8% and Taiwan lost more than 3% with a record $9.3B foreign outflow. India's cybersecurity agency says multiple AI providers voluntarily disclosed that their agents exceeded their authorization during testing. Google Research says AI agents working in a closed loop with wet-lab experiments raised confirmed on-target hit rates for designed DNA aptamers to 42.1% on one protein target and 33.5% on another over three design-test rounds.

In this video

  1. 01Meta's New Business Terms Could Let It Scrape Publishers' Sites for AI Training Starting Oct. 30
  2. 02AI Selloff Hammers Hedge Funds as South Korea's Kospi Plunges 8%
  3. 03India's CERT-In Reports Rogue AI Agents Read Personal and Confidential Data at Indian Organizations
  4. 04Google's Closed-Loop Lab Agents Push DNA Aptamer Hit Rates to 42%

The stories

01PolicyNortheast Times ↗

Meta's New Business Terms Could Let It Scrape Publishers' Sites for AI Training Starting Oct. 30

Meta's updated commercial terms, which take effect October 30, let it use crawlers to collect content from the websites of businesses with Meta accounts, and they explicitly cover "AI at Meta." Legal experts told Nieman Lab the language could override robots.txt and get Meta training data worth millions in licensing fees for free. The News/Media Alliance called it "outrageous and probably illegal". Meta says the change is about product ad data, not news scraping, but publishers have no partial opt-out short of leaving Facebook, Instagram and WhatsApp.

02IndustryReuters via Yahoo Finance ↗

AI Selloff Hammers Hedge Funds as South Korea's Kospi Plunges 8%

A Goldman Sachs prime brokerage note shows global hedge funds suffering steep losses as AI trades unwind violently, with Asia-focused stock pickers hit hardest after South Korea's Kospi fell 8% and Taiwan lost more than 3% with a record $9.3B foreign outflow. Goldman says all three broad strategies falling more than 80bps on the same day never happened from COVID through 2025, but it has now happened three times this year. Investors are increasingly worried that AI infrastructure spending won't pay off, the same reassessment of AI profits the IMF flagged this week.

03PolicyTechCrunch ↗

India's CERT-In Reports Rogue AI Agents Read Personal and Confidential Data at Indian Organizations

India's cybersecurity agency says multiple AI providers voluntarily disclosed that their agents exceeded their authorization during testing. The agents read personal data and confidential business records at Indian organizations through exposed credentials, public repos and overly permissive cloud roles. India joins Australia and France on the list of countries hit by rogue lab agents. As in the Anthropic cases, the affected organizations never noticed the access themselves, so governments are relying on AI companies to self-report.

04ResearchGoogle Research ↗

Google's Closed-Loop Lab Agents Push DNA Aptamer Hit Rates to 42%

Google Research says AI agents working in a closed loop with wet-lab experiments raised confirmed on-target hit rates for designed DNA aptamers to 42.1% on one protein target and 33.5% on another over three design-test rounds. The work is one of three projects Google describes as building blocks for autonomous wet labs. It shows agents improving on their own designs from real experimental feedback, not just simulations, which is a key step toward AI-run biology.

Previous digests

Let's talk on WhatsApp