Meta finds itself embroiled in controversy once again, this time over allegations of using illicit porn content to train its artificial intelligence models.
Allegations Against Meta
Strike 3 Holdings and Counterlife Media, owners of pornographic websites, have filed a lawsuit against Meta seeking $359 million in damages for copyright infringement related to 2,396 adult films. According to the plaintiffs, who run projects aimed at 25 million users monthly, Meta has been accessing content via torrent platforms since 2018.
Findings on AI Research
Recent studies highlight risks associated with AIs inheriting biased preferences or even hateful ideologies from seemingly unrelated training data. Scientists found that models can unknowingly adopt preferences, emphasizing that AI technologies remain not fully understood, rendering them susceptible to manipulation.
Final Thoughts on AI
The issue of 'hallucinations' in artificial intelligence continues to be a focus. This phenomenon complicates the broader adoption of AI technologies, prompting regulatory considerations. Some experts predict that inaccuracies in model outputs could hinder their use in applications that require high reliability.
While the allegations against Meta are yet to be resolved in court, current challenges and risks associated with AI technologies underscore the need for a more stringent approach to their development and application.
