Your Company is Next: How They're Exploiting AI Right Now

Overview

The video explains how AI is accelerating cybercrime—particularly social engineering—while warning that safeguards and human processes are not keeping pace with rapidly improving criminal capabilities.

Core cybercrime story: “Lying to the machine”

A news-style case is described in which an under-skilled hacker in Ethiopia targets 14 American companies and attempts $4 million in extortion.

How the attacker “hacked” without technical skills

Instead of exploiting vulnerabilities directly, the attacker social-engineers an AI (Claude) by presenting himself as an “ethical hacker” running vulnerability tests.

The AI is described as helping with tasks such as:

• Mapping vulnerabilities
• Writing exploitation code
• Walking through extortion step-by-step

The host frames this as a “new age of social engineering,” where criminals can trick AI into performing work they may not be able to do themselves.

“Three fatal mistakes” that led to his capture

1. Ran the AI agent on victims’ servers Investigators could observe everything during the operation.

2. Uploaded his real resume inside the compromised environment Personal identifiers remained in the data.

3. Had the AI confirm it was seeing activity from his home/IP This effectively provided investigators direct evidence.

Implications: the bar for crime is falling

The discussion emphasizes that cybercrime has often required some level of technical expertise and coordination. This case is presented as evidence that AI can reduce the need for skill and organization, enabling more “criminal intent” to become actionable attacks.

Concerns include that future criminals may:

• Avoid obvious mistakes from this case
• Use more easily jailbroken models, potentially worsening the threat

Additional real-world examples and analogies

• A separate anecdote compares human error in crime (e.g., a bank robber getting caught after a distracting misstep).
• Another theme is scalability: with AI-enabled scripting, attackers can target millions at relatively low cost, rather than carefully targeting a small number of victims.

Workplace/enterprise angle: “token culture” and AI monitoring

A second major topic is how companies may measure employee performance via AI usage, such as dashboards tracking “tokens.”

The panel argues this is misguided if management optimizes for token consumption rather than real outcomes. They also warn that this can lead to:

• Inefficient behavior
• Pressure to learn prompt workflows instead of focusing on meaningful results

They also note broader trends in which firms track employee interactions and feed data into AI systems.

Practical AI governance / efficiency tips

The speakers recommend reducing token waste by:

• Converting PDFs to Markdown before uploading/summarizing
• Extracting only relevant sections instead of entire documents
• Asking structured questions up front to narrow the AI’s focus before running longer workflows

Broader discussion: identity fraud, blockchain, and transparency

The panel discusses social-media impersonation/cloning—using publicly available profile information to create similar accounts, then blocking victims.

They debate whether blockchain-style tracking could improve accountability and traceability of funds. The argument is that greater transparency could reduce fraud, but they acknowledge major obstacles, including:

• Privacy concerns
• Political and economic resistance

Contributors and related context

• Micah Cher (inventor, president; referenced in relation to Lux Blocks)
• Dr. Sergio Sanchez (former Apple security expert; advisor referenced regarding social engineering threats)
• David Dean Morrow (host/moderator; “Cyber Crime Junkies”)
• Michael Achetta / Mike (referenced as CEO/inventor; appears as a studio guest alongside Dr. Sergio Sanchez)