Google, Microsoft, OpenAI, and Anthropic are forming an industry body called the Frontier Model Forum to advance AI safety research and responsible AI development. They have appointed Chris Meserole as the executive director and allocated a funding round of $10 million to red team the latest AI models. However, critics are skeptical about the fund’s ability to provide a balanced perspective on AI development, considering the involvement of major tech companies and the lack of transparency in AI training data.
Google has announced new tools to help users fact-check images on social media. The tools include viewing an image’s history, metadata, and context used on different sites. Users can also see when Google Search first saw the image and how it was described on other sites. Approved journalists and fact-checkers will be able to upload or copy URLs of images to learn more about them.
Nightshade is a new tool developed by Professor Ben Zhao and his team at the University of Chicago to help artists protect their work from AI image bots. By adding undetectable pixels that corrupt an AI’s training data, Nightshade alters the output of machine-learning models, making them interpret prompts differently. This tool, along with Glaze, another tool released by Zhao’s team, aims to encourage proper compensation for artists and deter copyright infringement by major AI companies.
OpenAI’s ChatGPT has combined its different abilities into a single chat, allowing users to upload and analyze various types of documents, browse the web, perform data analysis, and generate images using OpenAI’s image generation model DALL-E 3. Previously, users had to toggle each capability independently, but now they can use all these tools in the same chat session, greatly improving efficiency. OpenAI has not officially announced this bundled version of ChatGPT yet.
Let’s note that Google, Microsoft, OpenAI and Anthropic together only allocated ten million dollars for safety. Want to know why they aren’t trusted here on AI? You have only ten million answers.
With misinformation a chronic problem, tools like the ones described here may become essential for corporate communications and brand protection. MSPs could integrate these into their suite of services to add value and strengthen their cybersecurity posture, and help customers with the use of the tools themselves.