Sam Altman’s OpenAI has released GPT-4, an upgrade of its third-generation artificial intelligence engine that will affect the entire crypto industry.
The prior version, GPT-3, underpinned the viral hit ChatGPT which has millions of users and gained untold media exposure. GPT-3 also enabled Altman to raise a staggering $10 billion from Microsoft, valuing OpenAI as the world’s largest AI company.
Computer enthusiasts by trade, the crypto industry immediately heralded the launch of GPT-4. NFT fans immediately began experimenting with its art tools.
However, others cautioned about threats to blockchains and protocols with limited security testing, such as decentralized finance protocols.
Crypto flash loans vulnerable to GPT-4 exploitation
Many pointed to this week’s $197 million hack of Euler Finance, a new DeFi platform. Euler lost its reserves to an attacker who manipulated so-called “flash loans.”
- An attacker who takes out a flash loan trades using loaned funds and repays that loan — all within a single block on Ethereum’s blockchain.
- Flash loans allow any market participant to make large amounts of money using just a small amount of capital. The attacker need only post a small amount of collateral to leverage up to capitalize a major trade.
- Worse, flash loan-based attacks reward patient, diligent attackers who carefully plan their entire attack prior to executing any of its segments. This uniquely favors AI and Big Data tools that can comb through troves of data and suggest flash loan attacks whose entire sequence of events can be executed within a single on-chain block.
Once a flash loan attack begins, its profit is often guaranteed. For example, the Euler Finance attacker used malicious code to create artificially overvalued collateral and withdrew the real proceeds of a liquidation using that collateral.
In any case, the entire crypto community will be watching OpenAI’s new GPT-4 for any immediate effects on the digital asset ecosystem. Although permissionless financial exchanges like DEXs and other DeFi tools face the most immediate risks, the new AI will certainly influence NFTs, blockchain experiments, coding tools, security research, and other cryptographic initiatives.