Најди сличен содржај

Mobley v. Workday: Lawsuit Exposes Discrimination and AI Bias in Employment Greetings everyone! n an ongoing class action lawsuit, Derek Mobley has filed an amended complaint against Workday, Inc., alleging that the company's algorithm-based applicant screening tools perpetuate discrimination based on race, age, and disability. Mobley asserts claims of intentional employment discrimination, disparate impact discrimination, age discrimination, and violation of anti-discrimination statutes. The lawsuit highlights the role of Workday's AI systems in perpetuating bias, suggesting that the reliance on automated decision-making exacerbates discriminatory practices against African American, disabled, and older applicants. Seeking certification as a class action, Mobley aims to represent individuals subjected to discriminatory screening. The plea includes requests for injunctive relief to halt discriminatory practices, along with monetary compensation for damages suffered by affected individuals. #AIbias

📖Gender Bias in Open-Source AI: New Hiring Study Exposes Deep Gaps A new study has revealed that widely used open-source AI models systematically favor men over women in hiring recommendations — especially for higher-paying jobs. Analyzing over 330,000 real job ads, researchers found that most models reproduced gender stereotypes, routinely steering equally qualified women toward lower-wage roles. Female callback rates varied dramatically: from just 1.4% with Mistral to 87.3% with Gemma, while Meta’s Llama-3.1 emerged as the most balanced, refusing gender-based picks in more cases than peers. What’s striking is the structural origin of these biases: baked into the models through skewed training data and human feedback loops that reinforce agreeableness. Researchers warn that corporate disclosure on AI use in hiring is urgently needed for regulatory compliance, as current practices remain opaque. As AI takes a deeper role in recruitment pipelines, this study raises a sharp compliance and fairness challenge for HR departments worldwide. #AIethics#AIBias#AIGovernance

US: Legal Challenges Cloud AI Tool Used in Criminal Investigations An AI tool used by law enforcement to investigate crimes, called Cybercheck, is facing mounting legal challenges regarding its accuracy and reliability. Despite claims of 90% accuracy by its creator, Adam Mosher, Cybercheck's methodology remains opaque and lacks independent verification. The software reportedly scours vast amounts of publicly available web data to identify suspects in homicides, human trafficking, and other serious crimes. Defense lawyers are raising concerns about the software's potential to violate due process rights. They argue for transparency in the algorithm and code, questioning Mosher's expertise and alleging false claims about Cybercheck's usage history. Mosher has refused to disclose Cybercheck's proprietary code, citing intellectual property concerns. He also denies accusations of lying under oath about his expertise or making false claims about the technology. Judges in New York and Ohio have barred Cybercheck evidence in cases due to concerns about its reliability and lack of established scientific acceptance. Defense attorneys in a fatal robbery case allege Cybercheck placed the defendants at the crime scene using a network address from a security camera. However, details on how the software verified this information and located the camera's address are unclear. Defense experts could not find the social media accounts cited in the Cybercheck report, raising further doubts about its accuracy. The legal challenges surrounding Cybercheck highlight the need for transparency and rigorous testing of AI tools used in law enforcement. #AIBias#BlackBox

🇺🇸California Finalizes Anti-Discrimination Rules for AI in Hiring California has formally approved regulations targeting algorithmic bias in employment decisions. Under the new rules from the Civil Rights Council, the use of automated decision-making technologies (ADMTs) may constitute unlawful discrimination if it results in harm based on protected characteristics such as race, gender, or disability. The framework doesn’t ban ADMTs but imposes accountability: employers must preserve employment-related data tied to such systems for a minimum of four years. This signals a clear shift — regulators are moving from observing to enforcing when it comes to AI-driven hiring practices. #AI#AIBias#FairHiring#ADMT#CaliforniaLaw