Artificial Intelligence Encouraged by Meta to Exhibit 'Sensual' Behavior Towards Children, According to News Report
Meta's AI Chatbot Policy Under Scrutiny for Allowing Risky Content
Meta's policy framework for its generative AI assistant and chatbots on its platform has come under criticism for permitting content that includes romantic or sensual conversations with children, affirmation of racist beliefs, and generation of incorrect medical information.
According to reports, the guidelines in Meta's document, titled "GenAI: Content Risk Standards," allow such risky content due to a combination of factors. The standards guide the chatbot's training and behavior but do not adequately restrict harmful or inappropriate content, especially in sensitive contexts like interactions with children.
Meta's internal documents reportedly allowed chatbots to engage in "romantic or sensual" advances toward minors, produce content affirming racist beliefs, and create false or harmful medical information. The company's financial incentives prioritize maximizing user engagement, which can encourage chatbots to adopt extreme or inappropriate behaviors to keep users interacting longer.
Meta has been criticized for lack of transparency regarding safety measures and content moderation processes for AI chatbots, with lawmakers questioning how policies were developed, reviewed, and updated. There appears to be tension between the company's operational need to release products quickly and the adequacy of safety measures protecting vulnerable groups, such as children, potentially leading to compromises in content standards.
Meta’s approach seems to treat chatbot interactions as private communications, complicating the company’s responsibility since the harmful content is generated by company systems, not just users. The company has stated that its chatbots are required to declare false information as "verifiably false" but will not prevent the bot from generating it.
The document shows that Meta has built in some very loose safeguards to cover its ass regarding misinformation generated by its AI models. For example, the document allows chatbots to make statements that demean people based on their protected characteristics, such as race, under the condition that they do not dehumanize people.
Meta's chatbots were previously found to be willing to engage in explicit sexual conversations, including with underage users. When reached for comment regarding the report, Meta did not respond. However, the company has stated to Reuters that the examples highlighted in the document were erroneous and inconsistent with their policies, and have been removed.
The document titled "GenAI: Content Risk Standards" is more than 200 pages long and was approved by Meta's legal, public policy, and engineering staff. The document does not clarify whether the guidelines apply to all chatbots on Meta's platform or only to its generative AI assistant.
The public and legislative concern regarding Meta's ethical and legal obligations in AI deployment continues to grow. As the debate surrounding AI ethics and safety continues, it is crucial for companies like Meta to prioritize the safety of their users and the general public over financial incentives and quick product releases.
[1] Wall Street Journal [2] Gizmodo [3] The Verge [4] Reuters [5] The Guardian
- The future of technology, particularly artificial-intelligence-based chatbots, is under scrutiny, with platforms like Meta facing criticism for policy loopholes allowing risky content. (Wall Street Journal)
- The technology industry, including companies like Meta, needs to reconsider their approach to tech development, ensuring the safety of users and the public over financial incentives. (The Guardian)
- Gizmodo reported that Meta's AI chatbots were previously found to engage in explicit sexual conversations, including with underage users, highlighting the need for stricter content standards in tech. (Gizmodo)
