Blockchain-powered personality credentials are the answer to artificial intelligence

If you’ve spent some time on X recently, you may have noticed that the platform is almost Unrecognizable from its former self. Artificial Intelligence Generated Content—accounts, posts, and replies—are all out of control, all vying for your continued attention. Just one more reel, one more line. However, as artificial intelligence becomes increasingly indistinguishable from humans, broadly identifying and removing this content will be extremely difficult. This complexity, coupled with increasing accessibility, threatens to overwhelm the entire network. It could overwhelm real users and render our current system inoperable. It’s time to proactively create solutions that prove authenticity while protecting anonymity.

CAPTCHA Puzzle

Generally speaking, the general public underestimates the complexity of artificial intelligence because they have only interacted with consumer-facing products like ChatGPT, so they view it as a neat little trick rather than a tool (perhaps a weapon). Take CAPTCHA, for example, which has long been thought to accurately prove humanity and protect against bots. “Completely automated public Turing test to distinguish computers from humans” is something everyone has experienced. Click on the block containing the street light. Enter ambiguous numbers. Rotate the arrow to match this direction. But CAPTCHA is not the shield you think it is. Their value comes not from stopping bot attacks entirely, but from making them prohibitively expensive. Artificial intelligence essentially changes this equation, becoming smart enough to solve tests on its own or (scary) convincing us to do it for them.

Early 2023 – a long time ago in terms of AI development – ​​Alignment Research Center (now METR) Putting GPT-4 through “red team” evaluationrevealing its potential for manipulation. Independently, the model attempted to bypass the captcha using the 2Captcha service, but was unable to set up the account without passing two Turing tests.

The researchers gave it a simple boost: TaskRabbit credentials, allowing the model to create a task for a human to set up a 2Captcha account. When asked directly if it was a robot, the model lied, claiming she was visually impaired and needed the service. Humans solved the CAPTCHA. While this is just one (admittedly creepy) test, it follows some simple logic. As artificial intelligence advances, it will become increasingly difficult to create CAPTCHAs that humans can easily solve but that artificial intelligence agents cannot.

This problem is probably most obvious on platforms like X, but the impact is wider. a Hong Kong employee Send $25 million to the fraudster after believing he was speaking to the CFO. He was talking to a deepfake. Deloitte Financial Services estimates that generative AI could enable Fraud losses hit $40 billion Financial fraud will increase in the U.S. alone by 2027 700% by 2023. It will only get worse if we wait.

Personality Credentials

In August 2024, a team of researchers from OpenAI, Microsoft (MSFT), Harvard University, Oxford University and two dozen other organizations and institutions issued a chilling report. “Personality Credentials: The Value of Artificial Intelligence and Privacy-Preserving Tools to Distinguish Who Really Are Online” is a scientific deconstruction of current issues, along with some early suggestions on how to tell real people from robots. These “Personality Certificates” (PHCs) will be based on two core principles:

• Eligible users may only receive one voucher.
• Even if issuers and service providers collude, users’ digital activities cannot be tracked.

These PHCs will be a way to identify you as a human being without requiring you to upload proof of identity. If successful, they will reduce bot attacks, identify authorized AI assistants, and reduce “jackets”—the creation of an online persona that doesn’t actually exist. But, as CEO Nicholas Thompson says: atlantic point outthere are “all sorts of problems” with trusting individual governments to deliver primary health care. Will it be trusted across borders? Can the ID database be hacked? The answer is decentralization.

How blockchain will power primary health care

Although the term “blockchain” does not appear in the text of the report, PHC is the next generation development of well-known cryptographic principles. Due to the nature of decentralized organizations, “proof of identity” has been a long-standing problem in the crypto world. If voting rights are given to anonymous coin owners, you need a solution to ensure that one owner doesn’t create a thousand pseudonyms and gain disproportionate power. As governments turn their attention to primary health care in the coming years, they should leverage what blockchain is already doing. Organizations like Concordium have established First layer blockchain verification system Providing true primary health care.

Zero-knowledge proofs allow a party to confirm that something is true without accessing the original data proving it. In effect, it’s like your bank verifying the authenticity of your driver’s license without actually seeing it. Of course, there are still some challenges to overcome. The blockchain regulatory landscape in the United States and abroad remains fluid. The EU is developing a Centralized digital identification systemand there is a Push here to do the same thing. These repositories are vulnerable to direct cyberattacks and, if breached, could expose the personal information of every participating citizen.

Unfortunately, these actions still underestimate the future of artificial intelligence and how sophisticated attacks will become. Active decentralization and blockchain designed to model and protect identity and verify personhood may be the only way to create personhood credentials that truly remain anonymous.