Bybit: AI saves $700 million annually after being hacked by $1.46 billion

Bybit's $1.46 billion vulnerability set a reference point for centralized exchange risk, but the company now views AI as a tool to avoid further deterioration in bad years. The exchange said the systems had recovered $700 million in losses, a figure that provides a solid basis for the security arguments Bitcoin developers have repeatedly emphasized for weeks.

This claim, originally reported by CoinDesk, is not about the dollar amount itself, but about who is making the claim. Bybit is the first large centralized exchange to no longer limit itself to vague statements of AI monitoring after suffering one of the largest thefts in cryptocurrency history, but to announce specific savings.

The timing is not accidental. North Korean hacking groups remain the most persistent operational threat to exchanges, and the industry as a whole has struggled to prove that the money invested in detection and transaction screening produces measurable protection after an intrusion. A $700 million amount-even if unaudited-would be enough to shift the discussion away from theoretical capabilities to claims of actual results.

Calculation logic behind the numbers

Bybit did not disclose a specific calculation method for US$700 million. That alone is enough to keep analysts cautious. Estimates of savings in the security sector are often based on "possible losses" rather than hard accounting records. Still, the size is reasonable when you take into account hours-long withdrawal freezes, asset recovery efforts, and the portion of transaction costs that AI models may flag before funds are transferred.

Since being hacked, the exchange has been rebuilding its infrastructure. AI-based monitoring of suspicious withdrawal patterns, address blacklisting and real-time anomaly detection have become standard topics in centralized exchanges. The difference is that Bybit is now willing to attach a specific amount, which sets expectations for future disclosures by other operators.

This shift has had an impact on how exchanges market their compliance plans. Risk managers have long wanted to tie safe investments to reductions in expected losses; traders rarely see these numbers. Bybit's disclosures have prompted a more standardized approach, even if such standards have not yet been developed.

This is also driving a broader wave of AI in the cryptocurrency space. Projects are combining decentralized computing power with AI workloads, from scalable AI-driven Web3 applications to storage networks-where AI storage needs have become part of the price narrative. Security budgets are also part of this capital flow, although they receive less attention than consumer-facing AI products.

Unresolved issues

No external auditor verified the $700 million figure, and Bybit's statement did not clearly define the calculation cycle or method, making it impossible for traders to compare it to industry benchmarks. This gap is significant. Without consistent benchmarks, AI savings statements can become marketing indicators rather than risk indicators.

In addition, there is tension between the exchange's security narrative and unresolved threats from North Korean organizations. Previous thefts have shown that state-backed participants are able to transfer assets through core exchange infrastructure, rather than just targeting individual user accounts. If the control failure occurs during the internal transfer process, AI transaction monitoring can only partially repair and cannot replace more powerful key management and manual authorization control.

Regulators are also focusing on the same issue. As Washington continues to discuss options for structuring the country's largest cryptocurrency market, the legislative battle over exchange regulation shows how the security failure of centralized exchanges affects broader policy debates. A public AI savings figure can be used both to demonstrate that exchanges can effectively self-regulate, and to require stricter custody standards.

Exchanges face tougher issues

Bybit does not claim that AI could have prevented the initial hacking attack. The $700 million involved post-event operational defense. This difference is important. The industry needs to know whether these tools can deter sophisticated nationally backed attackers or mainly reduce the small-scale fraud that typically occurs after major breaches.

The more important test for users is whether deposit and withdrawal controls have become stricter without slowing down normal transactions. Exchanges that over-correct may push trading volume towards platforms with less censorship, creating another risk.

The coming months will show whether other exchanges will try to follow Bybit's disclosures and release their own numbers. If they do, the market will eventually have a benchmark set against which to compare AI security investment. If they don't, Bybit's numbers will remain an isolated data point that can win headlines but not enough to quell the debate about whether AI can significantly improve the security of exchanges.