AI NewsWaveGenon Launches Enterprise AI Guardrail Solution GenGuard
Key points
- Genon launches AI guardrail "GenGuard"
- Evaluates requests and responses against corporate security policy
- Built on experience running its own platform "GenOS"
Genon (CEO Ko Seok-tae) said on the 26th that it has launched "GenGuard," an AI guardrail solution that responds to security threats to generative AI services and manages security policies for each company. AI Times said it "has launched" it, while iNews24 and Techworld News said it "will launch" it.
Digital Today explained that GenGuard evaluates and blocks requests entered into AI services and the responses they generate according to security policies set by the company, and said in its headline that security policies can be set in natural language. Hellot reported that it applies company-specific security policies to respond to personal data leaks and prompt attacks.
As for the background to its development, AI Times wrote that Genon developed the standalone solution based on its experience running the guardrail function of its own generative AI platform, "GenOS." Digital Daily and iNews24 also described the solution in connection with the existing GenOS platform.
The market backdrop was also mentioned. Techworld News, citing Genon, pointed out that AI-enabled cyberattacks have recently been increasing. Hellot described the launch as a move to expand the business into the enterprise AI security market.
Genon said earlier, on the 21st, that it passed mass-production verification in an AI manufacturing automation project in Indonesia and then won follow-up projects. According to iNews24, Genon began work last year on a project for AI vision inspection of outsoles for the global sports brand Nike. VentureSquare reported that the technology recorded defect detection rates of 99.7% and 99.2% on two product styles.
Glossary An AI guardrail is a safeguard that keeps generative AI from receiving dangerous or inappropriate input or producing such output. A prompt attack is an attack that manipulates the text entered into an AI to induce it to break its set rules or leak internal information.