Here's what you'll find in this fortnight's AI updates:
FTC initiates inquiry into AI investments and partnerships for BigTech
White House calls for action after Biden and Taylor Swift deepfakes surface
Google introduces Lumiere AI, a text-to-video diffusion model
Draft Guidelines releases for standardization of China’s AI Industry
Singapore proposes model AI governance framework for responsible Generative AI
Irish government approves ethical AI compliance for public service tools
Australia unveils interim response to AI Regulation Proposals
FTC INITIATES INQUIRY INTO AI INVESTMENTS AND PARTNERSHIPS FOR BIGTECH
On 25 January, 2024, the Federal Trade Commission announced that it intends to look into the investments and partnerships for generative AI entered into by 5 major BigTech companies, including Alphabet, Inc., Amazon, Inc., Anthropic PBC, Microsoft Corp., and OpenAI, Inc.
These inquiries were launched by issuing orders under Section 6(b) of the FTC Act, that require the aforementioned companies to disclose strategic rationale for entering into the partnerships or investments, decisions around new product releases, analytics on the competitive landscape, competition in the resources for AI model inputs, and information provided to local and foreign government agencies including for investigative purposes.
Over the last year the Chairperson of the FTC, Lina Khan has initiated action against several tech behemoths including Google, Amazon, and Meta, in an attempt to curb their alleged unethical data privacy practices, by urging AI companies to uphold their users' data privacy, fining several tech companies for alleged unauthorised sale of personal data, and countering anti-competitive practices.
WHITE HOUSE CALLS FOR ACTION AFTER BIDEN AND TAYLOR SWIFT DEEPFAKES SURFACE
On 27 January 2024, the White House issued a statement urging the need for immediate action by Congress for legislating AI, in light of President Biden's voice clone calling for voters to cast ballots in November instead of January, and Taylor Swift's pornographic deepfakes circulating on the social media platform, X (formerly, Twitter).
X provided assurance that they were working on taking down the deepfakes involving Taylor Swift, and blocked all searches of her name on their platform, while the New Hampshire attorney general's office launched an investigation into President Biden's vocal spoof.
GOOGLE INTRODUCES LUMIERE AI, A TEXT-TO-VIDEO DIFFUSION MODEL
On 24 January 2024, Google introduced Lumiere, its text-to-video diffusion model designed to synthesize videos depicting realistic, diverse, and continuous motion, which was otherwise a major challenge in video synthesis.
Features of this model include text-to-video and image-to-video diffusion, stylised generation that creates videos in similar style as the reference, stylisation of videos based on the prompt, cinematography, and video inpainting that completes masked portions of videos based on their contexts.
DRAFT GUIDELINES RELEASED FOR STANDARDIZATION OF CHINA’S AI INDUSTRY
China’s Industry and IT Ministries released a draft guidance to standardize China’s AI industry on 17 January 2024. The Guidelines call for the widespread adoption of several standards including common standards such as AI terminology, architecture, and evaluation, support standards like computing devices, data services, and computing centres, key technologies pertaining to various machine learning methods, intelligent products and services like robotics and automations, industry applications covering manufacturing and multi-level adoption of AI, and security and governance.
The draft proposes the creation of more than 50 national and industry-wide standards, participation in setting more than 20 international standards for AI, and setting a milestone of achieving 60% standardised outcomes in industrial and project applications, all by 2026.
SINGAPORE PROPOSES MODEL AI GOVERNANCE FRAMEWORK FOR RESPONSIBLE GENERATIVE AI
On January 16, 2024, Singapore’s AI Verify Foundation and Infocomm Media Development Agency (IMDA) proposed a Model AI Governance Framework for Generative AI. It addresses risks such as bias, misuse, hallucination, and aims to provide a strategy to address them. The framework emphasizes nine dimensions, which include accountability, data, trusted development and deployment, incident reporting, resting and assurance, security, content provenance, safety and alignment, and AI for public good.
IRISH GOVERNMENT APPROVES ETHICAL AI COMPLIANCE FOR PUBLIC SERVICE TOOLS
The Irish Government, on 9 January, 2024, approved guidance for compliance with seven ethical AI requirements outlined in the European Commission’s guidelines for AI tools used in public service. The government emphasized seizing AI opportunities responsibly, supporting public servants in navigating ethical considerations, and adapting to the evolving landscape for the benefit of the served people.
AUSTRALIA UNVEILS INTERIM RESPONSE TO AI REGULATION PROPOSALS
On January 17, 2024, the Australian government responds to community input on AI regulation, confirming plans to create a framework for “high-risk” AI applications. The definition of "high risk" remains pending but is likely to involve systemic, irreversible, or perpetual consequences. Proposed actions include exploring mandatory safeguards, collaborating with the National AI Centre to develop an "AI Safety Standard," and considering voluntary labeling for AI-generated content in high-risk scenarios. The response aims to build on existing laws, reforms, and international commitments while emphasizing continued investment in national AI and robotics capabilities.
Authors: Keshav Singh Rathore, Astha Singh, Shruti Gupta.