Artificial intelligence agents are integrated throughout decentralized finance (DeFi) as the automation of financial systems and trading strategies trends within crypto, but how safe are they?

AI agents rely on precise data to operate effectively, automating liquidity provisioning, trades and portfolio management. Still, questions regarding their safety and reliability remain vital as adoption grows.

Mike Cahill, CEO at Douro Labs and contributor to the Pyth Network, told Cointelegraph in an exclusive interview that these AI agents rely on “real-time, high fidelity data to make split decisions.”

Errors or manipulation in the data could lead to unintended decisions with significant consequences, which Cahill said can be addressed by providing “ultra-low-latency, first-party price updates.”

Related: AI agents’ market cap surges 222% in Q4 2024, driven by Solana

Real-time data in AI decision-making

AI agents depend upon precise and timely data to make effective decisions in fast-moving markets, ideally sourced directly from first-party providers like market makers or exchanges. 

These systems aggregate data from multiple sources, reducing the likelihood of manipulation or inaccuracies, with sub-second updates as fast as one millisecond, ensuring agents can respond instantly to market changes.

“Pyth ensures agents operate on the most accurate market data available —eliminating risks tied to stale or manipulated information,” Cahill said. “AI agents thrive on speed, precision and automation.”

Related: ai16z’s Eliza Labs ships white paper for Web3-native AI agents

Safeguarding AI agents 

One of the biggest challenges for AI agents is operating safely within volatile market conditions, an element that decentralized systems aim to safeguard against.

For example, Oracle Integrity Staking (OIS) requires data publishers to stake capital, aligning their financial incentives with the accuracy of their contributions — meaning they lose their stakes if faulty or manipulated data is provided.

Cahill said that Pyth’s OIS creates an “economic security layer” that aligns with the first-party price sourcing and weighted aggregation to develop resilient, high-frequency pricing that reflects true market conditions. He added:

“AI agents can also integrate programmable safeguards, such as confidence intervals and predefined slippage thresholds, preventing them from executing trades under volatile or unreliable conditions.”

Related: Fetch.ai launches $10M accelerator for AI agent startups

Securing AI agents in DeFi

Cahill envisions a future where “fully autonomous financial systems operate more efficiently than any human-run market ever could” and expects to see artificial general intelligence (AGI) agents within the next one to three years.

“Real-time data gives AI agents the ability to unlock a new era of high-frequency, algorithmic trading in DeFi,” the Pyth contributor said. “This is where institutional DeFi surpasses TradFi, offering a market that’s not only decentralized but also faster, more efficient and truly autonomous.” 

The push for AI agent technology can be seen in recent developments in the crypto industry, such as the Fetch.ai launch of its $10 million accelerator for AI agent startups and the developer behind ai16z publishing a white paper on its vision for Web3-native AI agents.

Magazine: They solved crypto’s janky UX problem. You just haven’t noticed yet