Google prepares deepfake killer. Know how it will do it

Over the past couple of weeks, India has seen some ways in which deepfakes can be deeply disturbing and how the government of India is hell-bent on curbing the problem that comes out of AI-generated deepfakes and false information.

Google, a global tech giant, has provided insights into its efforts to counter deep fakes and AI-generated misinformation in India. Acknowledging the absence of a “silver bullet” to combat these challenges, Google outlined its strategy involving a combination of machine learning and human reviewers to identify and flag AI-generated content.

In collaboration with the Indian government, Google intends to address safety and security risks associated with ‘synthetic media,’ including deep fakes and disinformation campaigns, during the upcoming Global Partnership on Artificial Intelligence (GPAI) Summit.

To tackle the problem of fake images, Google has implemented guardrails and safeguards, including SynthID.

This system, an embedded watermark and metadata labelling solution, flags photos generated using Google’s text-to-image generator, Imagen.

Additionally, the company employs machine learning alongside human reviewers to swiftly detect and remove content violating guidelines, enhancing the accuracy of its content moderation systems.

For YouTube, Google is introducing disclosure requirements for creators using altered or AI-generated content. Creators will be mandated to inform users by adding labels to the description panel and video player.

The platform is also developing a ‘privacy request process’ enabling users to take down content that utilizes AI to imitate an individual’s face or voice.

In response to the evolving landscape, Google has updated its election advertising policies. Publishers are now required to declare if their ads include digitally altered or generated content with the intent to deceive, mislead, or defraud users.

Google Search incorporates guardrails such as Knowledge Panels and Featured Snippets to flag deep fakes and AI-modified content.

Engaging with policymakers, researchers, and experts across India, Google emphasizes its commitment to responsible AI practices.

In December of the previous year, the company allocated $1 million in grants to the Indian Institute of Technology, Madras, to establish a Responsible AI centre, focusing on studying bias in AI from an Indian perspective.