OpenAI Tests Zero Retention as Anthropic Moves to 30-Day Data Storage
Synopsis
OpenAI is testing a zero-retention safety system designed to protect customer data while detecting misuse, intensifying its privacy battle with Anthropic.
OpenAI zero retention has become a new battleground between OpenAI and Anthropic as both companies offer different mechanisms to ensure the retention of business data while addressing hypothetical security issues.
OpenAI is testing a private safety processing system that would allow the company to analyze patterns across a subset of interactions while keeping customer conversations private. The private safety processing test is only available to organizations and API customers and is not available to Individual ChatGPT Users.
OpenAI tests privacy-focused safety system
The zero retention system would allow the OpenAI system to flag down conversations that might require follow up, with no customer conversations being processed through the OpenAI system, except those that triggered a safety signal.
The company provides zero data retention on a specific API for qualifying enterprise users, and its enterprise privacy policy states that customer input and outputs are typically retained for 30 days on a system for the purpose of service or abuse analytics, and qualifying customers can request the ability to opt-in to zero data retention.
OpenAI has been developing procedures that will give the customer control of the infrastructure and encryption keys to prevent the company from accessing the business’s information, according to the new system report.
Anthropic Takes a Different Route
Meanwhile, Anthropic has pursued a different route in regards to data retention on its covered models, announcing the creation of an individual retention policy, which requires the retention of data on some prompts and outputs for 30 days for security reasons.
Anthropic’s data centers utilize this retention policy for Claude Mythos 5 and Claude Fable 5 models, according to Anthropic’s documentation, with data being deleted 30 days later unless its safety or legal teams determine otherwise.
With the increased use of artificial intelligence (AI) in corporations, safety has become a growing concern. The use of generative AI by the world’s workforce hit 17.8 percent during the latest quarter, according to the Global AI Diffusion Report.
The U.A.E. has the highest percentage of its workforce using AI services at an extraordinary 70.1 percent, followed by Singapore 63.4 percent, Norway 48.6 percent, Ireland 48.4 percent, and France 47.8 percent. The U.S. ranked 21st in the list with 31.3 percent. The situation with safety is particularly delicate for corporations in Australia and America, where trade secrets can be classified as critical corporate assets.
Corporations are especially concerned with protecting their client, business, and financial data, and the new security protocol is designed to help them regain confidence in the big spend on the combat as the revenue from the competition was $6.7 billion for OpenAI in its latest recorded quarter but operating losses rose to $12.3 billion.
Source: TechCrunch
Pooja Malik is a business journalist with over six years of experience covering startups, entrepreneurship, and emerging trends. She has previously worked with leading media platforms such as YourStory Media and BW BusinessWorld, where she reported on business, policy, and market developments. Currently, she serves as Editor at The Inspirepreneur Magazine, where she writes and edits stories across business, lifestyle, and travel, with a focus on clarity, accuracy, and reader relevance.
