Bloomberg Television12 min
How AI is Accelerating Cyberattacks
Courtney Drake, Anand Oswal, Anne Cleveland, Heath Fountain
Other
Surge in AI-Enhanced Cybercrime:
- Cyberattacks are occurring faster and at a larger scale due to AI integration by bad actors, increasing risks to businesses, infrastructure, and critical systems.
- FBI data reports over one million reported cybercrimes in 2025, with numbers rising sharply since 2020.
- 56% of attacks reported by organizations in the last year involved AI, representing a significant year-over-year increase.
- Over one in four organizations indicate AI was involved in the attack against them.
- Critical infrastructure targets include U.S. water utilities and European supermarket refrigeration systems.
- High-profile individual cases include Courtney Drake, a small business owner who suffered 460,000 unauthorized payments totaling over $1.2 million between June 26 and July 6.
- Drake incurred fees from Stripe on 500 processed fraudulent charges, in addition to dispute fees.
Asymmetry Between Attackers and Defenders:
- AI lowers the barrier to entry for attackers while increasing the volume and sophistication of threats.
- Palo Alto Networks researcher Anand Oswal notes defenders must be correct every time, whereas attackers need to succeed only once.
- AI allows attackers to scan open-source packages and identify vulnerabilities at an accelerated rate; for instance, Palo Alto Networks found roughly 10,000 vulnerabilities in two weeks that previously took years to discover.
- The concept of "security by obscurity" is now obsolete, as even small organizations and previously hidden bugs are prime targets.
- Technological debt accumulated over decades is currently coming due, exacerbating vulnerabilities.
Disproportionate Impact on Small Entities:
- The Center for Long-Term Cybersecurity (UC Berkeley) highlights that small organizations (municipalities, utilities, schools, nonprofits) lack resources to defend against nation-state actors.
- Small business owners are forced to act as cybersecurity experts, diverting time from income-generating activities.
- Colony Bank CEO Heath Fountain notes AI acts as a "force multiplier" for smaller banks but simultaneously heightens security needs as fraud costs rise post-pandemic.
Financial Impact and Strategic Responses:
- IBM data indicates the average cost of a data breach is nearly $5 million, covering crisis management, operational disruption, and reputational damage.
- 85% of organizations plan to increase security spending to counter AI-driven threats.
- Palo Alto Networks' AI systems currently block 31 billion attacks every 24 hours, including 14 million previously unseen (net new) attacks.
- Data exfiltration can occur within minutes of compromise, necessitating AI-driven automated responses rather than manual intervention.
Evolution of Application Security:
- AI-powered applications represent the "third wave" of application development, involving complex interactions between models, infrastructure, tools, and data.
- This architecture significantly expands the attack surface compared to traditional microservices or cloud-native applications.
- The fundamental cybersecurity risk formula remains: Motivation × Capability.
Governance and Human-Centric Solutions:
- Experts argue cybersecurity is primarily a risk and governance problem, not just a technical one.
- Leadership involvement is critical; CEOs and boards must govern cybersecurity risk, and similarly, mayors and principals must oversee security for local governments and schools.
- Basic cyber hygiene remains effective, including multi-factor authentication, passkeys, and verifying identity changes via phone calls.
- Community banks like Colony Bank emphasize educating customers on simple verification steps to prevent fraud.
- Organizations with strong leadership governance are statistically more successful at implementing technical controls.
Forward-Looking Outlook:
- Experts warn that defense systems have fallen behind the pace of technological advancement.
- The central challenge is preventing "Pandora's mistake" by taking immediate action to ensure AI is used to contain, rather than open, the digital threat landscape.