The H.A.I.R AI Regulation Updates are sponsored by Warden AI.
Warden AI is the tech-led AI auditor that helps HR Tech platforms and HR/TA teams demonstrate AI fairness and compliance. Our platform continuously audits AI systems for bias and provides transparency with real-time dashboards and reports. We work with leading AI talent platforms like Beamery and enterprise HR teams to mitigate risk, build trust, and meet AI regulations like NYC LL 144 and the EU AI Act.
Council of the European Union adopts rules to protect digital platform workers
The Council of the European Union adopted rules to improve working conditions for digital platform employees in the EU. The new platform work directive will bring more transparency to algorithms used in human resource management by ensuring that automated systems are monitored by qualified workers and give digital platform workers the right to contest automated decisions.
LLMs capable of reading 'invisible text' in prompts
Remember the white text in a CV to 'trick' keyword matching in an ATS? Well, it turns out that Large language models can read "invisible characters" because of a component in the Unicode text encoding standard and combined with normal text that users unknowingly paste into prompts in a practice known as "ASCII smuggling," ARS Technica reports. The LLMs most susceptible to ASCII smuggling were the Claude web app and Claude API from Anthropic.
LinkedIn pauses AI training data collection in Hong Kong
Following on from various other countries around the world, Hong Kong's Office of the Privacy Commissioner for Personal Data welcomed LinkedIn's pause on using Hong Kong users' data to train its AI models while it works with the PCPD. The social media platform had a default opt-in setting users had to toggle off if they did not wish for their data to be used, which the PCPD flagged as a privacy concern.
AI service enabled users to create chatbot personas of anyone without consent
A Character.AI user created a chatbot using the web content of a deceased woman without her family's consent, Wired reports. The publication found "several instances of AI personas being created without a person’s consent, some of whom were women already facing harassment online."
Meta launches AI assistant in the UK
I've been looking forward to this for some time! The Guardian reports Meta launched its AI-powered assistant, which can be reached through its Facebook and Instagram apps or its new glasses, in the U.K. Meta AI is not available in the EU due to regulatory challenges, according to the company.