OpenAI tests watermarking for images in GPT-4o

8 April 2025 2 minutes Author: Newsman

OpenAI is testing a feature to watermark images created in GPT-4o. While the feature is still in testing stages, evidence of its existence has already been found in the code of the Android app. This could mean that free users will receive images with a logo, unlike premium users.

AI researcher Tibor Blaho found a reference to the function “image-gen-watermark-for-free” in the code, which directly indicates a limitation for free users of ChatGPT. Users with a ChatGPT Plus subscription will likely be able to upload images without watermarks.

In addition, Blaho revealed several other tools that the company is working on:

  • Structured Thoughts
  • Reasoning Recap
  • CoT Search Tool
  • l1239dk1 (functionality unknown yet)

OpenAI has also expanded its Student Plus referral program, adding students from Australia and Colombia. For the US and Canada, ChatGPT Plus for students remains free until the end of May.

It is important to note the changes to the terms of use that give OpenAI the right to reproduce, modify and publish images and videos uploaded by users to the platform, including the Sora explore page.

Since the launch of its own Image Generator in GPT-4o, OpenAI has been actively implementing new features, expanding the creative possibilities of users. However, this raises concerns about:

– protection of created content,
– transparency of authorship,
– new restrictions for free users.

The company has already received comments from media lawyers about the limits of permissible use of AI-generated images, especially in commercial and educational contexts.

The introduction of watermarks could be the next step towards monetizing visual content created with GPT-4o. This is a reminder that the terms of use of OpenAI’s services are changing, and users need to be vigilant for updates.

Other related articles
News
Read more
Fresh Sakura RAT Appears on GitHub
A new Sakura RAT has surfaced on GitHub, demonstrating its ability to bypass antivirus and EDR systems through sophisticated obfuscation techniques including HVNC, injection, and registry persistence. This remote access Trojan is becoming a serious threat to corporate security, especially as similar tools become more widely available in the open source community.
196
News
Read more
Suspected Chinese hackers exploit new vulnerability in Ivanti firewalls
Hackers from China are exploiting a new weakness CVE-2025-22457 in Ivanti firewall systems, including Connect Secure and Policy Secure, used by government agencies. Despite the availability of the patch, attacks continue, and the company advises to completely abandon outdated devices. This confirms the growing threat of state cyberespionage through edge devices.
151
News
Read more
Port of Seattle reports data breach for 90,000 people after Rhysida attack
Following a cyberattack by the Rhysida group, the Port of Seattle confirmed that the personal data of about 90,000 people, including driver’s licenses and medical records, was compromised. Despite the disruption, key infrastructure services were not affected. The incident was one of the most high-profile examples of ransomware affecting U.S. government agencies.
148
Found an error?
If you find an error, take a screenshot and send it to the bot.