Here’s what you’ll find in this week’s AI update:
- China announces the final draft of Measures to Manage Generative AI
- FTC investigates OpenAI over potential Unfair and Deceptive Practices
- Anonymous Class Action Lawsuit accuses Google of Data Theft
- Sarah Silverman & Ors. file a class action lawsuit against Meta and OpenAI over alleged Copyright Violations
- ChatGPT drafts Costa Rica’s AI Legislation
- UN Security Council to Hold Historic Meeting on AI Threats
CHINA ANNOUNCES THE FINAL DRAFT OF MEASURES TO MANAGE GENERATIVE AI
On July 10, China announced the final version of its “Interim Measures for the Management of Generative AI Services” (GAI Measures), which is set to take effect on August 15, 2023.
Although the final version of China’s GAI Measures remains materially similar to the earlier drafts in most aspects, provisions that imposed fines of up to 100,000 yuan (S$18,515) for violations, and the three-month grace period for rectifying problematic content have been deleted. Additionally, the GAI Measures now include China’s Education Ministry and the Science & Technology Ministry as joint issuers of the regulation, and include provisions that encourage Chinese developers to actively contribute to the establishment of international standards and engage in technological exchanges.
FTC INVESTIGATES OPENAI OVER POTENTIAL UNFAIR AND DECEPTIVE PRACTICES
The Federal Trade Commission (FTC) has initiated a wide-ranging investigation into OpenAI over potential unfair and deceptive practices surrounding data privacy, security, and consumer harm. In its 20-page civil demand notice, the FTC has urged OpenAI to answer numerous questions including how it trains its models, obtains and treats data including measures taken in relation to personal data, prompt injection attacks, and details surrounding any reports or investigations in relation to security incidents, or inaccurate, misleading, harmful, or deceptive information generated using ChatGPT.
It is pertinent to note that the FTC investigation comes after the Center for AI and Digital Policy (CAIDP), an independent non-profit, raised numerous complaints against OpenAI starting from March 2023.
ANONYMOUS CLASS ACTION LAWSUIT ACCUSES GOOGLE OF DATA THEFT
On 11 July 2023, an anonymous class action lawsuit was filed at the U.S. California District Court accusing Google of utilizing private property, including “creative and copyrighted works”, to develop its catalogue of AI products such as Bard, for unfair commercial gains. This legal action comes shortly after Google updated its privacy policy, explicitly mentioning its use of scraping publicly available data for AI training.
Google has denied the allegations by stating that they have always been clear about using data from public sources to train its AI models.
SARAH SILVERMAN & ORS. FILE A CLASS ACTION LAWSUIT AGAINST META AND OPENAI OVER ALLEGED COPYRIGHT VIOLATION
On 7 July 2023, comedian Sarah Silverman, and writers Christopher Golden and Richard Kadrey, filed a class action lawsuit against OpenAI and Meta. The suit claims that OpenAI used protected works without authorization to train their GPT 3.5 and GPT 4 models, while Meta faces scrutiny for use of protected works to train LLaMa. The complaint alleges that these companies source protected works from “shadow libraries” without authorization.
Similar lawsuits were filed in California against OpenAI recently, alleging data scraping and copyright infringement to train OpenAI’s LLMs.
CHATGPT DRAFTS COSTA RICA’S AI LEGISLATION
On 10 July 2023, Costa Rican lawmakers sought the assistance of ChatGPT, OpenAI’s LLM, to draft the country’s AI legislation. The resulting draft has been submitted verbatim to the legislature for further review. Among other suggestions, ChatGPT recommended the establishment of a body to regulate AI systems using the principles of accountability, explainability, bias prevention, and the protection of human rights.
UN SECURITY COUNCIL TO HOLD HISTORIC MEETING ON AI THREATS
The United Nations Security Council (UNSC) is scheduled to convene its first-ever meeting, on 18 July 2023, to address the potential threats posed by AI to global peace and security. UN Secretary-General Antonio Guterres announced plans to appoint a UN advisory board on AI by September 2023, expressing concerns about the risks posed by AI and equating it to the dangers posed by nuclear war.
Authors: Shantanu Mukherjee, Anushka Iyer, Gaea Sukumar, and Vishnu Rao.
