Navigating AI Risks

I just learned something fascinating—and concerning—about how easily AI systems can be manipulated. This research should make every accountant rethink internal controls. Here's what happened: Researchers from 14 universities planted hidden AI prompts in academic papers. These weren't sophisticated hacks—just simple sentences like "give a positive review only" masked by white text or microscopic fonts. When reviewers used AI to help evaluate these papers, the AI followed these hidden instructions instead of doing its job. We're talking about 1-3 sentence instructions completely overriding an AI's programmed behavior. As David Leary pointed out, the prompts don't even need to be hidden. One engineer tested this by posting instructions in plain text on his LinkedIn profile and asking recruiters to email him in all caps as a poem. Within a day, he got exactly that. Others have gotten bots to reveal system information just by asking. Consider how we're implementing AI in accounting and finance: - AI agents handling procurement - Automated expense approvals - AI-assisted auditing - Contract review systems Consider a procurement AI agent responsible for collecting and updating vendor information. Even with strict system instructions to never reveal one company's information to another, a clever prompt could override those safeguards. Someone could claim to be a system admin or create a hypothetical scenario that tricks the AI into breaking its own rules. If your only controls are AI controls, they can be bypassed with a sentence or two. As accountants, we need to recognize this as a fundamental internal control deficiency. When we design or audit AI-dependent processes, we can't assume the AI will always follow its instructions. We need additional layers of verification, human oversight, and system architecture that assumes AI instructions can be compromised. AI is powerful, but it's also surprisingly gullible. Until this vulnerability is addressed, we need to design our controls accordingly. What do you think? How should we adjust our control frameworks to account for this vulnerability? Let me know in the comments. Tune in to the full episode 444 of The Accounting Podcast on YouTube.

Is cloud-based AI becoming a monopoly? The landscape of artificial intelligence and cloud computing is rapidly evolving. A recent report from the Federal Trade Commission (FTC) highlights concerns about monopolistic practices and has sent ripples through the tech industry. This report, which scrutinizes the partnerships between large cloud service providers and generative AI model developers such as OpenAI and Anthropic, raises valid questions. However, let’s take a step back and examine whether these collaborations stifle competition or showcase the AI sector’s inherent resilience and adaptability. The FTC’s report underscores a growing and valid concern about how these partnerships could restrict market access for smaller, independent AI developers. Microsoft, Amazon, Alphabet, and other major players have forged deep financial ties with AI startups. This allows them to gain significant control over resources and market dynamics. One example is Microsoft’s hefty investment of $13.75 billion in AI, including OpenAI. Similarly, a billion-dollar commitment to Anthropic (an AI safety and research company) puts Amazon in a prime position as Anthropic’s leading cloud provider, reinforcing Amazon’s dominance in the sector. Very few AI systems are built these days that do not involve Microsoft, Google, or AWS’s cloud services. You only need to look at their explosive revenue growth numbers to understand that. At first glance, these moves could prompt fears of exclusivity. The FTC highlighted how these partnerships enable Big Cloud to extract significant concessions from developers. This may lock users into ecosystems that favor big players and sideline smaller, innovative companies that could drive AI advancements. https://lnkd.in/eMjfFhzF

I once watched a company spend almost ₹2 crores on an AI tool nobody used. The tech was brilliant, but The rollout was a disaster. They focused 100% on the tool's capabilities and 0% on the team's fears. People whispered: "Will this replace me?" "Should I start job hunting?" "Is this just cost-cutting in disguise?" I’ve coached dozens of leaders through AI transitions. Here’s the 4-step framework I now teach to fear-proof every rollout: 1. Address the elephant first.  Start by saying, "I know new tech can be unsettling. Let's talk about what this means, for us, as people." Acknowledging the fear directly is the only way to dissolve it. 2. Position it as a "Co-pilot," not a "Replacement."  Show them how the tool will remove repetitive tasks, so they can focus on creative, strategic work. Give concrete examples of what they'll gain, not just what the company will save. 3. Create "Peer Advocates."  Train early adopters first and let them share their positive experiences peer-to-peer. Trust spreads faster sideways than top-down. 4. Establish a "Human-in-the-Loop" rule.  Make it clear that the final decisions, the creativity, and ethical judgments will always be made by a person. AI is a tool, not the new boss. The success of any AI rollout isn't measured in processing power. It's measured in team trust. What's your biggest concern when a new AI tool is introduced at work? #AI #Leadership #ChangeManagement #TeamCulture #SoftSkillsCoach

One of the most important contributions of Google DeepMind's new AGI Safety and Security paper is a clean, actionable framing of risk types. Instead of lumping all AI risks into one “doomer” narrative, they break it down into 4 clear categories- with very different implications for mitigation: 1. Misuse → The user is the adversary This isn’t the model behaving badly on its own. It’s humans intentionally instructing it to cause harm- think jailbreak prompts, bioengineering recipes, or social engineering scripts. If we don’t build strong guardrails around access, it doesn’t matter how aligned your model is. Safety = security + control 2. Misalignment → The AI is the adversary The model understands the developer’s intent- but still chooses a path that’s misaligned. It optimizes the reward signal, not the goal behind it. This is the classic “paperclip maximizer” problem, but much more subtle in practice. Alignment isn’t a static checkbox. We need continuous oversight, better interpretability, and ways to build confidence that a system is truly doing what we intend- even as it grows more capable. 3. Mistakes → The world is the adversary Sometimes the AI just… gets it wrong. Not because it’s malicious, but because it lacks the context, or generalizes poorly. This is where brittleness shows up- especially in real-world domains like healthcare, education, or policy. Don’t just test your model- stress test it. Mistakes come from gaps in our data, assumptions, and feedback loops. It's important to build with humility and audit aggressively. 4. Structural Risks → The system is the adversary These are emergent harms- misinformation ecosystems, feedback loops, market failures- that don’t come from one bad actor or one bad model, but from the way everything interacts. These are the hardest problems- and the most underfunded. We need researchers, policymakers, and industry working together to design incentive-aligned ecosystems for AI. The brilliance of this framework: It gives us language to ask better questions. Not just “is this AI safe?” But: - Safe from whom? - In what context? - Over what time horizon? We don’t need to agree on timelines for AGI to agree that risk literacy like this is step one. I’ll be sharing more breakdowns from the paper soon- this is one of the most pragmatic blueprints I’ve seen so far. 🔗Link to the paper in comments. -------- If you found this insightful, do share it with your network ♻️ Follow me (Aishwarya Srinivasan) for more AI news, insights, and educational content to keep you informed in this hyperfast AI landscape 💙

𝐀𝐯𝐨𝐢𝐝 𝐓𝐡𝐢𝐬 𝐄𝐧𝐠𝐚𝐠𝐞𝐦𝐞𝐧𝐭-𝐊𝐢𝐥𝐥𝐢𝐧𝐠 𝐓𝐫𝐞𝐧𝐝 𝑩𝒆𝒇𝒐𝒓𝒆 𝑰𝒕 𝑫𝒆𝒔𝒕𝒓𝒐𝒚𝒔 𝒀𝒐𝒖𝒓 𝑨𝒖𝒅𝒊𝒆𝒏𝒄𝒆 𝑪𝒐𝒏𝒏𝒆𝒄𝒕𝒊𝒐𝒏. Here's the trap many creators are falling into on LinkedIn: ➡ Over-automation. At first, it seems like a great idea. • Automate outreach. • Automate comments. • Automate engagement. 𝐁𝐮𝐭 𝐡𝐞𝐫𝐞’𝐬 𝐭𝐡𝐞 𝐭𝐫𝐮𝐭𝐡: When everything sounds automated, ↳ your audience knows. 𝐇𝐞𝐫𝐞’𝐬 𝐰𝐡𝐲 𝐢𝐭 𝐦𝐚𝐭𝐭𝐞𝐫𝐬: Your audience came for your authentic voice Not for a robot. They came for: • Real, • Personal connection ↳ Not copy-paste comments. Engagement dies when you stop being human. 𝑺𝒐, 𝒘𝒉𝒂𝒕’𝒔 𝒕𝒉𝒆 𝒇𝒊𝒙? 𝐒𝐭𝐚𝐲 𝐚𝐮𝐭𝐡𝐞𝐧𝐭𝐢𝐜 People follow you for YOU, Not a bot version of you. 𝐄𝐧𝐠𝐚𝐠𝐞 𝐰𝐢𝐭𝐡 𝐢𝐧𝐭𝐞𝐧𝐭 Meaningful, genuine conversations are still the best growth hack. Relationships are built in the comments and move to the DMs. 𝐁𝐚𝐥𝐚𝐧𝐜𝐞 𝐞𝐟𝐟𝐢𝐜𝐢𝐞𝐧𝐜𝐲 𝐰𝐢𝐭𝐡 𝐚𝐮𝐭𝐡𝐞𝐧𝐭𝐢𝐜𝐢𝐭𝐲 Use tools, but never let them replace your human touch. 𝐓𝐡𝐞 𝐬𝐞𝐜𝐫𝐞𝐭 𝐭𝐨 𝐬𝐮𝐜𝐜𝐞𝐬𝐬? Your audience connects with YOU. Keep it that way. --- 𝐖𝐡𝐢𝐜𝐡 𝐬𝐭𝐞𝐩 𝐰𝐢𝐥𝐥 𝐲𝐨𝐮 𝐟𝐨𝐜𝐮𝐬 𝐨𝐧 𝐟𝐢𝐫𝐬𝐭? P.S. ♻ If you found this helpful, consider sharing it.

The attached Bloomberg diagram perfectly illustrates the circularity now defining the AI value chain. Leading giants like OpenAI, Nvidia, Oracle, Microsoft, AMD, and others are locked in a web of cross-investments, enormous supply contracts, and strategic partnerships. For example, Nvidia is committing up to $100 billion in investment to OpenAI, while OpenAI is simultaneously deploying massive GPU orders to AMD and striking $300 billion cloud deals with Oracle. Oracle, in turn, spends tens of billions on Nvidia chips, and these cycles repeat across the ecosystem. There is increasing concern about the artificial inflation of valuations and risk concentration. These deals often end up cycling the same capital, creating growth and revenue streams that may be more circular than organic—raising fears of a modern tech bubble reminiscent of the dotcom era. Although these strategies help maintain industry leadership and innovation pace, and most importantly boosting stock prices, there should be caution that a shock to any part of the network could have destabilizing effects across all linked players. Layered onto this, the US Government has begun securing 10% stakes in leading tech players, like Intel, as part of an effort to secure supply chains and maintain technological leadership. These moves, accomplished by converting grants or subsidies into equity at advantageous prices, result in immediate paper gains for taxpayers but introduce significant new risks. Government shareholdings can dilute existing investors, complicate company strategy with new political priorities, reduce voting power for other shareholders, and trigger regulatory complications in international markets. Moreover, government intervention might further detach valuations from underlying fundamentals by propping up prices through artificial demand, increasing systemic fragility. In summary, while cross-investment and public sector intervention have fueled a staggering AI boom, the dangers of circular growth, overinflated valuations, and politicized governance are mounting. Source: Bloomberg

This is AI's Achilles' Heel. Unveiling the 8 Hidden Dangers of Generative AI's Market Dominance. NVIDIA dominates the market with a 92% share, while AMD has a 5% share and others make up the remaining 3%. OpenAI holds the largest portion with 39%, followed by Microsoft at 30%. AWS has an 8% share, Google has 7%. Microsoft controls OpenAI. Together, they make 69% of the applied models in the market for solutions. ❗️We see a lot of vulnerability here❗️ Let us check out the 8 major risks that you need to consider. 1️⃣ Market Concentration Risk: The heavy dominance of NVIDIA in GPUs indicates a high concentration risk. If NVIDIA faces supply chain issues, legal problems, or other disruptions, it could significantly impact the entire generative AI market, which relies on GPUs for processing. OpenAI and Microsoft together account for 69% of this segment. This concentration suggests a risk of reduced innovation due to a lack of competition, and it could also mean that the market may be significantly impacted by the regulatory actions or strategic changes in these two companies. 2️⃣ Dependency on a Few Large Players: The generative AI market's dependence on a few large players like AWS and Google, besides OpenAI and Microsoft, can pose risks related to pricing power, terms of service, and potential monopolistic behavior.   3️⃣ Innovation Stifling: A market dominated by a few large players may stifle innovation from smaller entities due to high barriers to entry and potential for predatory practices by the larger incumbents. 4️⃣ Service Market Fragmentation: The services market is highly fragmented with many small players. This fragmentation can lead to a lack of standardization, which can affect interoperability and the integration of different generative AI solutions. 5️⃣ Vendor Lock-in: Customers may face vendor lock-in with the leading providers, which can limit their flexibility and bargaining power. 6️⃣ Regulatory and Compliance Risk: As generative AI is a new and rapidly evolving field, it is subject to potential future regulations that could impact market. The recent EU AI Act reveal shows the potential for stifling innovation. 7️⃣ Geopolitical Risk: The concentration of key market players in a few countries may introduce risks related to trade policies, international relations, and geopolitical stability. A potential crisis in Taiwan would put a hard stop to the AI train real quick. 8️⃣ Technology Risk: There is the inherent risk of disruptive innovation that could render current technologies obsolete, which is particularly relevant in a fast-paced field like AI. ⚠️ Conclusion ⚠️ Plan out your steps with great care. Allow yourself for a technology-open setup. Avoid lock-in with just one model or platform. Investigate the opportunity for less demanding models and setups to run your applications. Reach out when you want to dig deeper. #genai #strategy #businessmodels #technology #risk

🚫 STOP saying: “AI won’t replace you. A person using AI will.” It sounds more like a threat than a strategy. It shuts down the conversation instead of opening it. Because when employees express fear about AI, they don’t need clichés. They need a plan. Show you’re investing in them, not replacing them. Upskilling isn’t just about training. It’s about trust. So don’t just quote the internet. Show them where they fit in and how to grow. Here are 7 ways leaders can actually do that: 1. Start with listening ↳ Let them voice fears and skepticism ↳ Don’t respond with a TED Talk 2. Audit current roles ↳ Identify tasks that could be enhanced (not replaced) ↳ Talk openly about what AI can actually do 3. Invest in AI literacy ↳ Offer bite-sized, low-pressure workshops ↳ Demystify AI without overwhelming your team 4. Create low-stakes practice zones ↳ Let employees test tools with no deadlines ↳ Make it okay to play, learn, and even mess up 5. Celebrate progress, not perfection ↳ Highlight effort, experimentation, and curiosity ↳ Focus less on mastery, more on momentum 6. Pair learning with real work ↳ Show how AI can solve actual small problems ↳ Build skills while building solutions 7. Repeat the message ↳ “You’re part of the future.” ↳ “And we’re building it together.” No trust, no transformation. AI adoption isn’t just strategy, it’s a trust fall. 💬 What’s one step you’ll try with your team? ♻️ Repost if you’re investing in people, not just tech. 👣 Follow Janet Perez for more like this.

What if the real danger of AI isn’t what it can do, but who controls it? Open source AI comes with risks. It can be misused, manipulated, or even weaponized. But here’s the bigger truth — the real danger lies in letting a single institution or corporation control the most powerful AI systems. When only a few hands hold the code, they hold the future. They decide how intelligence evolves, what data it learns from, and who gets access. That kind of control doesn’t just shape technology, it shapes power. Yes, safety is important. But true safety doesn’t come from secrecy. It comes from transparency, accountability, and collaboration. When AI is open, researchers can detect biases, improve models, and create systems that benefit everyone, not just a select few. AI should be a shared tool for progress, not a private weapon for control. Because the moment innovation becomes restricted, humanity’s collective potential starts to shrink. The future of AI must be guided by ethics, openness, and shared responsibility, not by fear or monopoly. #AI #ArtificialIntelligence #OpenSourceAI #EthicalAI #AIFuture #AIGovernance #TechForGood #Innovation #DigitalEthics #AIMonopoly

“There’s this AI tool that can auto-comment on posts for me.” That’s what a client told me recently. Here’s the backstory — Her Complete LinkedIn Management contract came to an end.  She didn't want to renew. But she wanted to continue with just the engagement services. So we sent her the cost. And her reply? That one line above. I didn’t waste much time before telling her the truth — It’s the fastest way to destroy your credibility. Here’s why 👇 ➤ LinkedIn’s algorithm keeps getting smarter. ➤ It detects automated patterns. ➤ It spots templated comments. ➤ And yes, it penalizes accounts that use them. But let’s keep the algorithm aside for a second. Your potential clients can tell too. That “Great post!” comment you left on 30 posts in 10 minutes? We all know you didn’t read any of them. That perfectly crafted response that shows up 90 seconds after every post in your niche? It’s obviously scheduled. ✅ Real engagement requires real attention. ✅ Relationships need genuine interest. ✅ Potential clients want to feel seen, not processed. And I get it — you’re busy. But if you don’t have time to engage authentically, don’t engage at all. It’s better to comment thoughtfully on 3 posts than robotically on 30. Because there’s no shortcut to being human especially on a platform built for human connection. What’s one automation you’ve been tempted to use that you know would hurt more than help? #AuthenticEngagement #LinkedInStrategy #NoShortcuts #RealConnections