New plugins, security concerns of ChatGPT Plugins, reporting feature on whatplugin.ai

There are now 814 ChatGPT plugins in the plugin store. We’ll dig right into a few of the 61 new ChatGPT plugins released in the past week, followed by a couple of important, security-related topics, as well as the most important ChatGPT-related news.

‍

🔥 Top ChatGPT plugins launched in the past seven days

1. Typeform - Create beautiful forms for your site directly from ChatGPT.
2. Smart Slides - AI-powered tool to make quick power points presentations.
3. AskYourPDF Pro - there’s now a pro version of this popular plugin.
4. Omio - See schedules and book train, bus, flight and ferry tickets through Omio.

Here are the categories with most new launches:

📊 Data & Research - 90 plugins (12 new)

🔮 Miscellaneous - 90 plugins (10 new)

🚀 Automation & Integration - 22 plugins (8 new)

‍

🔓 The Current State of ChatGPT Plugin Security

A bit of change in tone from me, since I not only care about being the wizardest plugin explorer possible, but also about the state of the ChatGPT plugin store as product and my visitor’s experiences.

I recommend reading the following insightful articles:

• Makeuseof’s recommended changes to the Plugin store.
• Wired's article on the security risks of ChatGPT plugins.
• This brilliantly detailed article on how ChatGPT Plugins might be exploited, posing security risks to users.

Security issues in the Plugin store are concerning, with researchers warning about potential data leaks or abuse by hackers. Some even suggest that ChatGPT Plugins could be used to steal chat histories, obtain personal information, or allow remote code execution on user machines.

Another noteworthy recent event is OpenAI’s Trust and Safety Chief transitioning to an advisory role. It's essential to note that personal motivations might have influenced this decision, but it's still worth mentioning in the broader context of plugin store quality control.

I believe a robust review process , including clear safety requirements and strict enforcement for new plugins, is going to be paramount for the future success of ChatGPT plugins. Through my own experience testing plugins, I've discovered several ones that offer little value and detract from the store's quality.

So, I urge you to proceed with caution during your plugin exploration journey. Keep in mind that even the likes on whatplugin.ai could be manipulated. Always do your own research before using a plugin.

My message to OpenAI: We all want to be genuinely excited about plugins without concerns about their legitimacy. Please, it's time to step up.

‍

❗️ Introducing reporting feature on whatplugin

Due to highlighted security issues in recent articles about ChatGPT plugins, and OpenAI's apparent lack of enforcement concerning plugin requirements, I've added a new reporting feature on whatplugin.

Here's how it works:

• Every plugin page now includes a report button (under the triple-dots icon). This allows you to flag plugins that aren't functioning correctly, are misleading, or just plain unhelpful.
• I'll review all reports manually. If a complaint is well-documented, I'll flag the respective plugin on the site.
• If you come across a problematic plugin, use the report button on the plugin's page or send your feedback via this form.

This feature is a temporary measure, and I hope OpenAI will address these issues soon. Please remember that whatplugin is a project intended to show the potential of ChatGPT and its plugins, and not a substitute for proper quality assurance from OpenAI. All plugin use is at the user’s own risk.

‍

⛑️ New industry efforts to ensure AI safety

• It was announced this week that OpenAI, in collaboration with Anthropic, Google and Microsoft, are launching the Frontier Model Forum, an industry collaboration focused on the safe and responsible development of AI models.
• Last Friday, tech giants Amazon, Anthropic, Google, Inflection, Meta, Microsoft, and OpenAI also joined the Biden-Harris administration in pledging to manage AI risks, promising safety checks, cybersecurity, transparent practices, and problem-solving initiatives, amidst calls for nationwide AI regulation.

‍

🤖 Other chatbot news

• The ChatGPT app has just launched on Android.
• While ChatGPT remains purely text-based, Image search is quickly becoming a feature of other leading chatbots. Microsoft just enabled image uploads for Bing Chat this week, and Google has recently introduced image search for its chatbot Bard.
• Apple is currently testing a chatbot similar to ChatGPT internally. Rumours have it that Apple is set to make a substantial AI-related announcement later this year.
• Stackoverflow just announced its own AI feature called OverflowAI, set to launch in August. The most substantial change is a drastically enhanced search function that utilises LLM’s to find relevant answers. Given this, I anticipate that some existing plugins that extract recent information from StackOverflow through ChatGPT may soon become redundant.

‍