When we think of AI (Artificial Intelligence), ChatGPT or Microsoft Copilot is the first thought that comes to mind. That’s primarily because, for most of us, it’s the only interaction we have had with one of the most significant developments of the 21st century. But AI is a lot more!

In recent months, we have witnessed considerable improvements, Sora AI by OpenAI being a prime example. Subsequently, the number of people raising concerns has been on the rise. And it’s coming from those who have worked closely with the technology right from the start.

Geoffrey Hinton, known as the Godfather of AI, last year quit Google to warn the world about the risks posed by AI in the absence of stringent regulations.

Another ex-Google employee, Mustafa Suleyman, has expressed concern that AI could be used to create another pandemic-like situation, one much worse than Covid-19.

What did Microsoft’s President Brad Smith say about AI?

In an interview with EL PAIS, Smith’s response, when asked if AI could result in a large percentage of the workforce losing jobs, was rather reassuring.

Use this new technology to exercise better judgment, to be more creative, to develop ideas, to help with writing but not delegate or outsource thinking or writing to a machine. If we use it well, it can be an accelerator for people and the kinds of work they do.

This has been the stance of most experts in the field. AI is a tool, and it should be used as such to boost productivity and improve output!

Smith also emphasized adapting to the changing world of technology, utilizing the potential of AI, and employing it in whatever role you can. Harnessing the technology is the key!

Coming to the regulation brought in by the EU (European Union) for AI, Smith was optimistic that it would set a benchmark for safety and security standards and provide protection.

When we buy a carton of milk in the grocery store, we buy it not worrying about whether it’s safe to drink, because we know that there is a safety floor for the regulation of it. If this is, as I think it is, the most advanced technology on the planet, I don’t think it’s unreasonable to ask that it have at least as much safety regulation in place as we have for a carton of milk.

Smith emphasized a safety brake for AI. He had brought up the subject previously in the U.S. Senate.

We better have a way to slow down or turn off AI, especially if it’s controlling an automated system like critical infrastructure.

Another concern raised during the interview was how the use of deep fakes could affect us, especially in a year when roughly 64 countries are going to poll. Smith underscored the steps taken by Microsoft to tackle deepfake and the need to educate people.

The more powerful the technology becomes, the stronger the safeguards and controls need to become with it. I think that all of us are going to have to push ourselves. The biggest mistake the tech sector could make is to think that it’s already doing enough and it can just do what it needs to be done if it’s left alone.

Regarding countries using AI for conducting cyberattacks, Smith had an insightful response citing how AI can be used to counter such issues.

We’re not going to let nation state actors that engage in this kind of adversarial and harmful conduct have use of our applications because we regard that as something that is likely to do harm to the world. But we also need to use AI to fight back and create stronger cybersecurity protection.

Is AI really a threat?

Whether you believe it or not, AI is still in its infancy, and over the years, its integration with everyday tasks will increase even more. Surely, it will lead to job losses, but so does every technological advancement!

It’s rather early to discuss whether AI poses an existential threat to humanity. But there must be safeguards and regulations because there are other more pressing issues of the day that these can address. Be it deepfakes, cyberfrauds, or harassment!

Another concern is the biased training of AI models. At the time of writing, a recent example is Google’s Gemini image creator, which many reported had a bias. After a massive uproar, Google temporarily halted the AI-tools ability to generate images of people.

So, we must learn how to harness and better utilize AI and, at the same time, introduce safeguards to protect us from any unfavourable scenarios.

What do you think of AI and the threats posed by it? Share with our readers in the comments section.

Recently, Microsoft released an open automation framework, PyRIT (Python Risk Identification Toolkit for generative AI), which aims to empower security experts and ML engineers to identify risks and stop their gen-AI systems from going rogue.

The tool is already used by Microsoft’s AI Red Team to check risks in its generative AI systems, including Copilot.

In the announcement, the company emphasized that the release of the internal tool for the use of the general public and other investments made in red teaming AI is part of its commitment to democratizing AI security.

What is the need for Automation in AI Red Teaming?

Red teaming AI systems involves a complex multistep process, and Microsoft’s AI Red Team has a dedicated interdisciplinary group of security experts for this purpose.

However, Red-teaming generative AI systems are different from conventional AI systems; here are some major ones:

• Addresses security and responsible AI risks at the same time – The traditional AI systems mainly focus on determining security failures; however, the red-teaming generative AI systems check for both security risks and responsible AI risks, including inaccuracies, fairness issues, or content generation problems.
• Different architecture – The architecture of gen-AI systems varies from standalone apps to integration in existing apps with varied input & output modalities like audio, images, text, and videos.
• Probabilistic nature – Unlike classic AI systems, the same input can produce different outputs in gen-AI systems due to various factors like app-specific logic, orchestrator dynamics with extensibility, the variability of the generative AI model, and language input nuances.

This is why PyRIT was introduced, as it can address all the challenges posed by gen-AI red teaming.

How does PyRIT work?

First, PyRit Agent sends a malicious prompt to the Targeted Gen AI System; when it receives a response from the Gen AI system, it sends a response to the PyRIT scoring engine.

Next, the scoring engine sends the response to the PyRit Agent; then the agent sends a new prompt based on the feedback from the scoring engine.

This automation process continues until the security expert gets the desired result.

In a red teaming exercise, the Microsoft team picked up a harm category, generated thousands of malicious prompts, and used PyRIT’s scoring engine to evaluate the output from the Copilot system in a few hours instead of weeks using manual methods.

The results showed that using the tool can increase security experts’ efficiency in finding loopholes and preventing them from being exploited.

The Redmond tech giant also mentioned that PyRIT is not a replacement for manual red teaming but serves as a complement.

It augments the expertise of the security professionals, automates daily tasks, and makes them more efficient. At the same time, control of the strategy and execution of the AI red team operation always remains with the security experts.

PyRit consists of five components: Targets (supports various gen AI target formulations), Datasets (Allows encoding of static or dynamic prompts), Scoring Engine (scoring outputs using a classical machine learning classifier or an LLM endpoint), Attack Strategy (supports both single-turn and multi-turn attack strategies), and Memory (Saves intermediate input and output interactions for in-depth analysis).

You can download the toolkit from the GitHub website, and to get accustomed to it, the initial release includes demos, & common scenarios.

Have you tested your Gen AI system using PyRIT yet? If yes, share your experiences in the comments section below.