AI ethics and regulation took center stage in a recent conversation with former New York Manhattan District Attorney Cyrus Vance Jr. on the NYCLA Amicus Curiae podcast. Vance, now global chair of Baker McKenzie’s cybersecurity practice, discussed the technology’s potential, its pitfalls, and the role of both government and private firms in safeguarding the public.
From Vatican conferences to school curricula
Vance recalled attending a Vatican‑hosted conference on artificial intelligence, noting that the Pope’s interest in the topic “doesn’t surprise” him. He argued the Catholic Church, which runs the world’s largest non‑governmental school system, could develop curricula that teach children responsible AI use. “Millions of kids across cultures could learn about the risks and benefits,” he said, suggesting a coordinated educational effort could reach a global audience.
He emphasized that AI education should start early, covering basic concepts and safe practices. While no specific program was detailed, the idea reflects a broader trend of institutions seeking to embed digital literacy in school programs.
Benefits, biases and the “dark side”
Vance described AI as a “computer program” that processes vast amounts of data to generate responses word by word. He highlighted its usefulness in scientific research, medical diagnostics, and rapid information retrieval. “When a knowledgeable coder knows how to prompt the system, it can answer questions with impressive speed and relative accuracy,” he explained.
At the same time, the former DA warned that the technology is “value neutral” and that its impact depends on who controls it. He cited three areas where AI has already raised concerns: predictive policing, risk‑assessment tools such as the COMPASS system, and facial‑recognition software used by law enforcement. In each case, historical data can embed racial bias, leading to disproportionate outcomes for Black defendants.
Related: Exiting your firm Try a joint client notice
Vance also recounted a recent personal experiment with ChatGPT, which generated a professional biography for him based on publicly available information but mistakenly listed him as an “assistant district attorney.” The episode illustrates how AI can produce plausible but inaccurate content, a risk that extends to legal briefs and other official documents.
He described a “dark side” involving the creation of deep‑fake pornography featuring minors, as well as fabricated images of public figures. These examples highlight the growing challenge of “fake news” and malicious content that can spread quickly online.
From his prosecutor’s perspective, AI should never replace human judgment, especially in decisions that affect liberty. “A risk score can inform a judge’s decision, but it should not be the sole basis,” he said.
His view aligns with the European Union’s AI Act, which classifies law‑enforcement AI as high‑risk and requires conformity assessments before deployment. The EU legislation contrasts with the United States’ more permissive stance, where companies often launch products before regulators intervene.
In the middle of the discussion, Vance offered a practical take: for families living in neighborhoods where AI‑driven tools are already in use, the most immediate benefit comes from transparent explanations of how those tools work. When parents understand that a risk‑assessment algorithm draws on past arrest records, they can better question its fairness and demand oversight. This awareness empowers ordinary citizens to push for accountability without needing deep technical expertise.
Related: Birthstone Brilliance: Selecting a Gemstone Ring for Every Month
Regulatory recommendations and corporate responsibility
When asked what tech firms could do to curb abuses, the former prosecutor said they should treat AI products like any other consumer good. “You can’t sell a baby carriage without safety testing; the same principle should apply to AI,” he argued.
He noted recent state actions, such as New York and California laws that limit the hours children can access AI‑generated content and restrict the use of their data for targeted advertising. These measures represent the beginning of a “guardrails” approach that balances innovation with protection.
AI risks are real.
Vance concluded that while private companies have a role in self‑regulation, ultimate responsibility rests with lawmakers who must delineate permissible uses and enforce compliance. He warned that without such oversight, the line between beneficial innovation and harmful exploitation will remain blurred.
