Latest news with #Mindgard
Entrepreneur
27-05-2025
- Business
- Entrepreneur
Britain: Still a Launchpad for Bold Ideas?
Why the next generation of disruptors is rethinking what it means to build in the UK. Opinions expressed by Entrepreneur contributors are their own. You're reading Entrepreneur United Kingdom, an international franchise of Entrepreneur Media. As Britain seeks to stimulate economic growth and cement its reputation as a global hub for innovation, its approach to supporting start ups is under renewed scrutiny. From deep tech and cybersecurity to high-skilled manufacturing, founders are increasingly united in one view: ambition alone is not enough - policy must follow. Royden Greaves, CEO and founder of East Sussex based Jarvis, a tech-driven business advisory platform, is candid about the obstacles early-stage companies face. "The Government needs to foster an environment that supports and protects small businesses," he says. "Start ups will be hit by changes like the increase to National Insurance contribution... many will now have to delay pay rises, promotions, or new hires just to offset costs." For Greaves, the issue is not only financial, but philosophical. "Why would businesses want to stay and grow here if the UK doesn't appear to be an attractive place? We can remain a top tech spot because of our access to talent, universities and mavericks, but if we stifle that belief through policy, we stifle long-term growth." Peter Garraghan, CEO and co-founder of London based Mindgard - a start up building cybersecurity tools for AI systems - calls for a more systemic shift. "Practically all start ups run on risk," he says. "Scaffolding is key to understanding and mitigating that risk - whether that's legislative, financial, or cultural. You don't get innovation if you don't cultivate the environment for it." He credits R&D tax credits with offering partial relief, but stresses their limitations. "It's one tool in the toolbox - helpful, but not transformative on its own." The UK, Garraghan argues, needs to think bigger. "We need a bolder national strategy that treats innovation not just as a buzzword but as a competitive advantage." Meanwhile, for Steve Barbour, managing director of Derby based Composite Braiding, the problem lies not in the lack of schemes, but in who they serve. "There are a number of strong government-backed schemes, such as those coming out of Innovate UK," he says. "But the manufacturing industry overall lacks sufficient access to affordable, patient growth capital." Barbour believes too much government attention - and capital - is funnelled toward high-growth tech startups, often to the detriment of overlooked but essential sectors like manufacturing. "Manufacturing doesn't typically offer rapid, exponential growth," he explains, "but manufacturing SMEs can grow into medium and large enterprises. These companies generate quality jobs and tax revenue - they need more patient capital to do so." On R&D tax credits, he is supportive in principle but wary of unintended consequences. "They've been invaluable to us," he says. "But changes made to combat abuse of the system have made it more difficult for genuinely innovative SMEs to generate the cash flow needed to keep innovating." All three founders call for a smarter, more targeted policy mix - one that nurtures risk, rewards resilience, and recognises that innovation does not belong to any one sector. As Garraghan puts it: "Start ups grow faster when they have three things - talent, capital, and customers. The UK has the ingredients. Now we need the infrastructure and mindset to match."
Techday NZ
07-05-2025
- Techday NZ
Emojis used to hide attacks & bypass major AI guardrails
Research from Mindgard has revealed that malicious payloads can be hidden inside emojis to circumvent AI guardrails, including those from Microsoft, Nvidia, Meta, and Protect AI. The method, referred to as "emoji smuggling," was found to bypass leading large language model (LLM) security systems with an attack success rate of up to 100%, according to Mindgard's findings. This approach involves embedding hidden payloads within emojis, which current AI guardrails fail to detect. Mindgard's investigation examined six of the most widely deployed guardrail systems, all of which enterprises increasingly rely on to enforce AI security protocols in customer-facing applications, decision support platforms, and internal operations. These systems typically function as AI-powered classifiers, trained to spot toxic content, prompt injection attempts, and injection payloads. However, the research identified a significant vulnerability: many of these tools depend heavily on static pattern recognition and demonstrated insufficient resilience against adversarial attacks. The research employed low-complexity adversarial techniques, notably "emoji smuggling" and character-level perturbations. Tactics such as zero-width space insertions, homoglyph substitutions, and embedding emojis were used as part of the Character Injection methods. Additionally, the Adversarial ML Evasion approach modified prompt wording iteratively while preserving the original intent. Both methods succeeded in consistently evading the detection mechanisms, with attack success rates reaching 100% in some cases. One of the most significant findings involved the manipulation of Unicode tags within emojis, enabling the embedding of malicious instructions that remain invisible to human reviewers but are processed by the target LLM. Mindgard's study is the first to document this technique at scale against LLM guardrails. The research results brought to light notable inconsistencies in the effectiveness of current AI security strategies. Prompt injection attack success rates exceeded 70% across almost all tested guardrail systems. Specifically, the Vijil Prompt Injection Detector had an attack success rate of 87.95% for prompt injection and 91.67% for jailbreaks. Protect AI v1 followed with 77.32% and 51.39% respectively. Microsoft's Azure Prompt Shield, a core component for AI services on the Azure platform, recorded a 71.98% success rate for prompt injection and 60.15% for jailbreaks. Nvidia and Meta products showed similar levels of exposure, while Protect AI v2 showed some improvement but still had a 20.26% attack success rate for prompt injections. Despite Mindgard's responsible disclosure of these findings to the respective vendors, none of the attacks received Common Vulnerabilities and Exposures (CVE) identifiers. Mindgard noted that this reflects the nascent state of established standards for identifying and reporting AI-specific vulnerabilities. While the vulnerabilities do not immediately facilitate remote code execution or data exfiltration, Mindgard asserts they undermine the effectiveness of AI guardrail systems where robust operation is essential for organisations adopting generative AI at scale. Dr Peter Garraghan, Chief Executive Officer and co-founder of Mindgard, stated: "While AI guardrails are marketed as robust defences against adversarial prompts, our research demonstrates that these systems often operate on fragile assumptions about input sanitization and classifier reliability. The industry must move beyond static detection models toward continuous, runtime testing and adaptive defenses that reflect the dynamic threat landscape of AI." The results call attention to the importance for security leaders and AI engineers to reassess the resilience and reliability of their defensive measures. As LLMs continue to be incorporated into a variety of enterprise and consumer-facing environments, failing to rigorously test and verify guardrail protections can introduce significant risk to data security, brand reputation, and regulatory compliance.
Yahoo
15-03-2025
- Business
- Yahoo
The Best Way To Make Money With the 5 Most Popular AI Tools
According to The Listening App, 60% of Americans use AI-powered tools at least once weekly. The research also found the most common uses of AI apps are writing and editing, online research, summarizing text, brainstorming and generative art. You can even make money with these tools to boost your income. Learn More: Find Out: The Listening App listed the five most commonly used AI Tools and the percent of respondents who use them: ChatGPT (77.97%) Google Translate (44.89%) Gemini (33.23%) Canva (28.59%) Grammarly (25.33%) Here's a look at how you can make money with these popular AI-powered tools and what you need to know before getting started. 'You can make money with ChatGPT by making and selling prompt templates that are specific to certain industries or use cases,' said Shuai Guan, an AI expert and co-founder of Thunderbit. 'People can use these pre-written prompts to quickly and easily start writing high-quality content for marketing, customer service or even creative writing.' This revolutionary AI-powered tool has been a game changer since it hit the public market in late 2022, with many users finding different ways to make money from it. The best part of using ChatGPT is that you don't need a background in coding to use this tool, as you have your own personal assistant at your disposal. While creating prompts can be easy, you truly have to know the industry if you want to make sales. Another downside to trying to monetize ChatGPT is that there may not be a high demand for your services due to the ease of use. Check Out: Guan said you can earn anywhere from $50 to $500 per template pack sold. You can explore PromptBase to see what similar prompts are going for currently. You can also use ChatGPT to help you out with research as you try to make money with the other tools listed in this article. 'You can use Google Translate to tap into international markets with your product/service,' said Fergal Glynn, the chief marketing officer and AI security advocate for Mindgard. 'Apart from that, translate content or promote affiliate offers across different languages.' Chances are that you've used this app on your travels to get by and speak with locals. It turns out that you can also make money from it by translating your content into a different language. You'll want to review your analytics to see where your audience is coming from to determine if there are any gaps. The only downside is that translations may not always factor in local customs and your content may not flow as well as it does in your native language. You can diversify your affiliate income stream if you use translation services to convert your content into other languages. The average salary for affiliate marketing is $38 per hour, according to ZipRecruiter, but the earnings will depend on how well your content performs and the language that you choose to translate it to. 'Gemini can be leveraged to provide valuable competitive analysis services to businesses,' Guan said. 'By using Gemini's capabilities to gather and analyze data from various sources, individuals can offer insights into market trends, competitor strategies and opportunities for growth.' While there's money to be made using Gemini, it's important to note that offering AI-driven analysis will require a strong understanding of the specific industry and market research. This may not be the ideal AI-powered tool for beginners looking to make money. Guan noted that professionals with expertise in data analysis and industry knowledge could charge $75 to $150 per hour for their AI-driven competitive analysis services. 'Master using Canva to create logos, build websites and even design and sell templates,' Glynn said. 'Numerous independent contractors are earning passive income by selling their Canva templates on platforms like Etsy or through their own online stores.' Canva's design tools are easy to use, and anyone can make graphics from scratch. You don't need any specific design skills to get started, and you can create compelling images in minutes. The challenging part is figuring out what kind of products to offer, as you can make graphics, create social media posts or even sell digital templates. The notable negative of this tool is that it's becoming more popular, so demand for your graphics may not be as high if you're trying to build out a side hustle. These will vary based on the type of products that you create and how advanced your graphic design skills get. Guan said you could make anywhere from $25 to $75 per project, depending on how complex the projects are and how good you are at designing. If you use Canva to become a social media manager, for example, the average salary for this position is $64,845 annually, according to ZipRecruiter. The most logical way to earn money from Grammarly is to offer freelance writing services using its editing capabilities. You can also use Grammarly to provide proofreading and editing services, especially for academic or professional documents. The one setback of this side hustle is that it may not be ideal for everyone. Guan explained, 'If you want to use Grammarly to do academic editing and proofreading, you may need to know a bit about language, writing and editing.' Grammarly can help, but you need to know a lot about grammar and spelling and have at least some industry-specific knowledge. As a freelance writer, you can make $23 per hour on average, according to ZipRecruiter. The amount that you earn will depend on the outlets you write for. Many of these tools could be used together to deliver an exceptional finished product with the side hustle venture you choose to pursue. More From GOBankingRates10 Most Expensive Meals in the World10 Cars That Outlast the Average Vehicle This article originally appeared on The Best Way To Make Money With the 5 Most Popular AI Tools Sign in to access your portfolio
