Calling Out Commercial Motives Behind AI Safety Warnings
Zoho co-founder and chief scientist Sridhar Vembu has raised strong questions about major AI labs calling for industry-wide slowdowns while continuing to ship new models.
Speaking to ET AI, Vembu addressed recent calls from industry leaders—including Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman, and xAI chief Elon Musk—who advocated slowing development due to safety concerns. Pointing out that Anthropic and OpenAI launched new models shortly after these warnings, Vembu noted that any company genuinely concerned about risk has the autonomy to halt its own projects.
“They can slow it down themselves. They don’t need anybody’s permission,” Vembu told ET. “What they are saying is, let the others stop doing it. We will continue investing in it. That’s not acceptable.”
While not dismissing existential risk concerns entirely, Vembu urged an evidence-based approach rather than relying on commercial hype.
Prioritizing Productivity with Safety Over Raw Speed
Rather than chasing unchecked speed, Vembu revealed to ET that he devotes 70% of his time to making AI-generated software reliable and secure.
Comparing AI coding tools to high-speed trains, he emphasized that rapid generation requires robust tracks and signaling. Because AI-generated code can pass standard tests while harboring undetected security flaws or malware, human oversight remains non-negotiable. At Zoho, AI heavily assists programmers, but human developers retain full accountability for final software releases.
ZOHO'S AI DEPLOYMENT SAFEGUARDS
+-----------------------+-----------------------------------------------------+
| Human Accountability | Developers remain strictly responsible for releases |
| Code Security | Double-checking AI output beyond basic testing |
| HR Appraisals | Managers may use AI tools, but humans evaluate staff |
| Workforce Impact | AI assists support teams without displacing staff |
+-----------------------+-----------------------------------------------------+
Keeping Humans Accountable in Management and Upskilling
Vembu firmly rejected the idea of delegating sensitive human decisions to AI systems, telling ET that Zoho will not allow AI agents to conduct employee appraisals. While evaluators can use AI tools as assistants, human leaders must remain in charge.
Additionally, Vembu warned against over-reliance on AI tools among junior staff. While AI can offer short-term productivity gains, using it without understanding core fundamentals hinders long-term skill development. Companies must establish clear usage guidelines to ensure young professionals continue learning as the industry navigates this transition.
