OpenAI announces new AI safety program; researchers worry it won't be enough
OpenAI CEO Sam Altman said he would grant "employee level" access to outside researchers to evaluate AI safety. The pilot program appears much more restrictive.
Noticia publicada originalmente por NPR — News en npr.org (se abre en una pestaña nueva). Minuto 24 no es autor de este contenido.
Claves de la noticia
- OpenAI anunció un programa de evaluadores externos integrado en el que personal de cuatro organizaciones de seguridad en IA actuará como "embedded assessors".
- Investigadores de seguridad en IA sostienen que el programa se queda muy por debajo de la promesa de Sam Altman de septiembre sobre un acceso similar al de un empleado.
- Según los investigadores, los evaluadores tienen alcances muy limitados y solo dos parecen poder hablar libremente con empleados; no tienen acceso al Slack interno.
- OpenAI reconoce que todavía no existen estándares para las evaluaciones integradas y afirma que los evaluadores pueden pedir hablar con más empleados.
- Tres empleados de seguridad despedidos han expresado preocupación de que OpenAI esté dando marcha atrás en su compromiso con los evaluadores integrados.
Contexto
La seguridad de la inteligencia artificial se ha convertido en un tema central del sector, en parte porque los modelos son cada vez más potentes y sus posibles riesgos preocupan a investigadores, gobiernos y al público. Las evaluaciones externas, en las que especialistas independientes examinan los sistemas antes de que se usen ampliamente, son una de las principales herramientas para detectar fallos que las propias empresas podrían pasar por alto. Por eso, el grado de acceso que reciben esos evaluadores a los modelos, a los equipos y al personal es clave para juzgar si el control es real. El debate también se enmarca en un clima de atención pública tras incidentes recientes relacionados con sistemas de IA y en la competencia entre empresas como OpenAI y Anthropic por mostrar compromisos de seguridad.
Resumen elaborado por Minuto 24 con ayuda de inteligencia artificial a partir de la noticia publicada por NPR — News. Puede contener imprecisiones: ante cualquier duda, consulta la fuente original.

The OpenAI logo is displayed on a smartphone screen placed on a reflective surface onto which the company's logo is projected in Creteil, France, on Nov. 6, 2026. Samuel Boivin/NurPhoto via Getty Images hide caption
AI safety researchers say OpenAI's new third party evaluator program falls far short of what the company promised, raising concerns about its effectiveness.
The company announced Monday that staff from four AI safety and research organizations will become "embedded assessors" to "give people a clearer view of how labs are evaluating and managing [potential] risks."
"Getting an outside opinion on the safety work and the robustness of these systems is going to be increasingly important and valuable," CEO Sam Altman said at Semafor's Silicon Valley and the World conference on Monday.
However, many AI safety researchers say the new program is a far cry from Altman's vow in September that independent evaluators would have "employee-like access" to internal company systems.
Altman's promise came after details of the Hugging Face attack by OpenAI's autonomous systems shook the AI field. Worries of the technology's catastrophic risk captured public attention, and rival Anthropic said it would set up such a program.
"We are giving deep access to some of the most sensitive tools and models our researchers use...that are the most important for assessing key, relevant and timely safety claims," said Lama Ahmad, OpenAI's frontier assurances lead, on X.
Prior to these new embedded evaluation programs, AI companies have brought in outside researchers to assess the capabilities of AI models and identify safety loopholes the companies have missed.
The work of outside evaluators have already helped shed light on the companies' work, and have raised public awareness of safety risks that the technology poses. One recent report about Anthropic using AI to significantly speed up model development has prompted concerns in the field about how close the top companies are to using AI to autonomously build even more powerful models, a stage known as recursive self-improvement. Some in the field worry that if this happens, the velocity of development could lead to catastrophic consequences for humanity and are calling for new regulation against the practice.
Researchers say the actual announcements of how these programs will work suggests something more limited than what Altman promised.
"The assessors have very narrow scopes to their access, and only two of them seem to really be allowed to talk to employees," said Sydney Von Arx, founder of AI safety research group Nightingale.
One of the announced assessors, Apollo Research, outlined its limited access to OpenAI employees. If employees that haven't been whitelisted, or pre-approved, reach out to the embedded assessors, then the assessors will have to loop OpenAI in before having substantive conversations with the employees.
It's important for outside researchers to be able to interview employees, Von Arx said. "They help you find your blind spots. They tell you what you're missing."
Another glaring omission: "Nobody has access to internal Slack," Von Arx said, referring to the corporate messaging tool, which would allow assessors to see what employees are talking about with one another. "Imagine being an 'employee' and only being allowed to talk about company information with whitelisted coworkers and having no Slack."
In a statement to NPR, OpenAI said there's currently no standards for embedded evaluations. "We expect to deepen these relationships over time and share what we learn to help establish best practices for effective embedded assessments across the industry," OpenAI spokesperson Gaby Raila said. The company also said assessors can ask to talk to employees that haven't already been whitelisted, and that there is a process in place to get statements from OpenAI about things they've learned informally.
The third-party organizations will focus on various aspects of AI safety, according to OpenAI. Some of the work is related to preventing models from trying to engage in undesirable behavior, and other parts are related to catching bad behavior when it happens. The assessors will have access to some internal models, training processes and some OpenAI employees. The assessors will be able to publish their findings after OpenAI redacts confidential information, the company said.
OpenAI's commitment to embedded evaluations was questioned when three employees who worked on safety were fired. They subsequently said they were worried OpenAI is walking back from its promise of having embedded evaluators.
One of the fired employees, Tomek Korbak, was OpenAI's technical point of contact for two outside evaluators that investigated the Hugging Face attack.
"Embedded auditing will only fulfill its aims if the auditors really get employee-level access as opposed to being hired to do a specific job," he posted on X in response to OpenAI's announcement. "Sadly, OpenAI makes it clear now that [Sam Altman's] September tweet about 'employee-like access' was just an awkward overpromise that the company wants to quietly walk back."
NPR does not offer or accept money for coverage or interviews. Anyone who makes such an offer or request is not an NPR employee and does not represent NPR.



