AI safety and innovation were the focus when Dr. Chris Mattmann, Founder and President of Mattmann.AI, joined Kyra Phillips on ABC News Live. The October 5 conversation examined Sam Altman’s recent comments about accepting some negative consequences in exchange for artificial intelligence’s broader benefits, the federal government’s new “Superintelligence Force,” and whether voluntary industry safeguards have enough enforcement behind them.
The discussion came amid growing public concern about AI safety, data centers and the power technology companies have accumulated as artificial intelligence becomes integrated into more parts of society.
AI Safety and Innovation Are Not Mutually Exclusive
Phillips opened by asking whether society must allow bad things to happen before it can realize the benefits of AI.
Mattmann explained that Altman’s comments reflect a real tension. Advanced AI systems could be misused for cyberattacks and other dangerous purposes, particularly as agentic systems gain the ability to use tools and act more quickly than human defenders.
However, focusing exclusively on potential harm would ignore the enormous benefits AI could produce. Mattmann pointed to advances in mathematical reasoning, physics and fluid dynamics, along with the technology’s potential to contribute to new treatments for disease, new food sources and new forms of energy.
“We don’t have the guardrails yet, but there’s a lot of good too.”
The central policy challenge is therefore not choosing between technological progress and public safety. It is developing safeguards that allow society to pursue AI’s benefits without treating preventable harms as an unavoidable cost of innovation.
Agentic AI and Cybersecurity Risks
Mattmann emphasized that some AI risks are not merely theoretical. Models can already be adapted into cybersecurity agents capable of identifying vulnerabilities and executing tasks at a speed and scale that can be difficult for people to match.
This capability can strengthen cyber defense, helping organizations detect threats and respond to attacks. In the wrong hands—or without appropriate authorization limits—the same capabilities could also be used offensively.
Organizations deploying agentic AI therefore need strong access controls, continuous monitoring, clearly defined boundaries and meaningful human accountability. Giving an AI system tools and network access changes its risk profile substantially compared with a conventional chatbot that only generates text.
A Federal Home for Artificial Intelligence
The interview also addressed President Trump’s newly announced “Superintelligence Force” and the appointment of national intelligence leadership to a central role in federal AI policy.
Mattmann explained that the composition of the group demonstrates the growing connection between artificial intelligence and national security. It also creates an opportunity to address a persistent organizational question: where should responsibility for AI ultimately reside within the federal government?
Federal AI efforts have often been distributed across agencies, departments and separate initiatives. Mattmann said the new group should bring together technical talent, government expertise and industry partnerships while establishing clearer responsibility for how AI is acquired, governed and deployed.
Central coordination should not eliminate agency expertise. It should provide a common framework for accountability, safety and implementation across the government.
Do Voluntary AI Safeguards Have Enough Teeth?
Phillips also asked about a recent White House meeting with leaders from major AI and technology companies. Participants agreed in principle to safety measures, independent auditing and reviews of the alignment processes used to make advanced models behave consistently with human instructions and values.
Mattmann said those commitments represent a useful step, but the critical issue is enforceability. A voluntary or “morally binding” agreement is not equivalent to legislation, regulation or a formal compliance requirement.
If AI companies are primarily responsible for writing and enforcing their own rules, the public is left asking who evaluates compliance, what standards auditors must apply and what consequences exist when a company fails to meet its commitments.
Independent evaluation can improve transparency, but effective oversight also requires measurable standards, clearly assigned authority and consequences that extend beyond reputational pressure.
Building Trust Through Accountable AI Leadership
The ABC News Live discussion highlighted why AI safety and innovation must advance together. Artificial intelligence can accelerate scientific discovery, strengthen cybersecurity and support economic growth. It can also amplify malicious activity and concentrate significant power in a small number of organizations.
The goal should not be to stop AI development or accept avoidable harm as the price of progress. Government and industry leaders must establish safeguards that are technically credible, independently evaluated and enforceable.
Watch Chris Mattmann on ABC News Live
Watch the complete interview on ABC News.
Thank you to Kyra Phillips, Phoebe Gao and the ABC News Live team for the thoughtful conversation.
Mattmann.AI helps organizations connect artificial intelligence strategy and innovation with practical governance, safety and implementation. Learn more about Mattmann.AI.