In a significant announcement at the TED2025 conference, Sam Altman, the CEO of OpenAI, addressed the dual-edged nature of generative artificial intelligence (AI), emphasizing the urgent need to tackle privacy and security concerns that may hinder widespread adoption. Altman’s insights come at a time when tools like ChatGPT have seen unprecedented growth, yet many users remain hesitant to fully embrace this technology.
Altman outlined that while generative AI, particularly through platforms like ChatGPT, has rapidly gained traction—ChatGPT-4o, the latest image generator from OpenAI, reportedly gained one million users within just an hour of its launch—significant barriers still exist. "I think people will be slow to adapt to AI agents in many ways, but I also believe that even if some are comfortable and others are not, we will have AI systems that navigate the internet," Altman stated, highlighting the varying levels of comfort users have with AI.
He drew parallels between current hesitations towards AI and the initial reluctance many had towards online banking. "In the past, people were cautious about entering their financial data online until security systems evolved and trust was built gradually," he noted. This historical perspective underscores the need for robust security measures as AI systems become more integrated into everyday life.
Altman pointed out that the primary challenge facing AI today is security. The ability of intelligent systems to access personal computers and sensitive data significantly heightens the risks associated with potential errors. "When AI makes mistakes, the stakes are much higher because it has the capability to access your systems, your data, and even navigate through your device," he explained.
He concluded his remarks by asserting that safety is not merely a technological advantage but a fundamental basis for the success of AI products. "A good product is a safe product. People won’t use AI systems if they don’t trust that they won’t drain their bank accounts or delete their data," Altman stated, reinforcing the importance of user trust in the technology.
In a related development, OpenAI has recently launched a new feature in ChatGPT called "Return to Protected Memories." This innovative addition aims to enhance user interaction by allowing the AI to remember details from past conversations, thereby providing more personalized and effective responses. This feature represents a significant evolution of the previous memory capability, enabling the AI to retain information shared by users without requiring them to repeat it in future interactions.
The introduction of this memory feature is designed to make conversations with ChatGPT smoother and more accurate, bringing the AI closer to understanding human behavior. Users now have full control over how their memories are managed, with several options to maintain their privacy. They can completely disable the memories feature, manage or delete individual memories, and even activate a "temporary mode" that prevents any memories from being stored after a session ends.
OpenAI has assured users of its commitment to security and privacy, providing them with the freedom to decide what information is stored. "The new feature remains optional," the company emphasized, allowing users to choose whether they wish to utilize it. This approach ensures that conversations are not tracked or permanently stored unless users opt for greater personalization.
Despite the exciting advancements, the rollout of the "Return to Protected Memories" feature is not without its limitations. While it is available to ChatGPT Plus and Pro subscribers worldwide, there are regulatory constraints in regions such as the United Kingdom, European Union, Iceland, Norway, and Switzerland, where OpenAI is still reviewing compliance with local data protection laws. These reviews are crucial to ensure that the feature aligns with the legal frameworks in these areas.
As AI continues to evolve, the balance between innovation and user trust remains critical. Altman’s remarks at TED2025 and the introduction of the memory feature in ChatGPT reflect OpenAI’s ongoing efforts to address these concerns while enhancing user experience. By prioritizing security and privacy, the company aims to foster a safe environment for users to engage with AI technologies.
Ultimately, as generative AI tools become more integrated into daily life, the focus on building trust through robust security measures will be essential for their successful adoption. The journey towards a future where AI systems operate seamlessly alongside humans is still unfolding, and the path will undoubtedly require careful navigation of the challenges that lie ahead.
