AI Ethics Alarm: Anthropic's Claude Four Sparks Controversy as SMBs Navigate Economic Uncertainty

[00:00:02] It's Tuesday, May 27th, 2025 and I'm Dave Solt. Three things to know today. SMBs show cautious optimism with AI adoption still stumbling. Lenovo's earnings hit by tariffs spotlighting the current risks. Anthropics Claude Four sparks an ethical alarm bell with disturbing behaviors. And Washington eyes a decade-long freeze on state AI regulation while DHS bans commercial tools outright. We'll talk about why these matter.

[00:00:31] This is the Business of Tech. I hope everyone enjoyed the long weekend who had it, and it was both US and UK listeners for this one. So let's dive into the current market conditions. A new report from OnDeck in collaboration with Okrulus reveals that 93% of small business owners anticipate either significant or moderate growth in 2025, despite facing economic uncertainties and potential tariff changes.

[00:00:57] This reflects a slight decline from 94% in the previous quarter, indicating a cautious optimism among small businesses. The report highlights that over half of the surveyed businesses, specifically 55%, have already taken precautionary measures to prepare for potential disruptions, such as increasing inventory and securing additional capital.

[00:01:17] Furthermore, a notable shift in lending preferences is observed, with 76% of businesses now opting for non-bank lenders, reaching a new high in the survey's history. The findings are based on responses from 437 small businesses and reflect a growing trend towards digital transformation, with 69% reporting frequent use of artificial intelligence tools in marketing tasks. The full report is available on OnDeck's website.

[00:01:44] The latest report from the Bank of America Institute indicates a mixed outlook for small businesses, highlighting sector-specific trends in inventory management and hiring activity. While small businesses are generally not increasing their inventory levels, firms in the manufacturing and services sectors are ramping up spending to stock up in anticipation of potential disruptions.

[00:02:07] According to the report, a net negative 4% of business owners plan to invest in inventory in the coming months, contrasting with a 6% increase in hiring payments made above 2019 levels. However, this increase may reflect difficulties in filling roles, as 34% of small business owners reported unfilled job openings in April, the lowest figure since January of 2021.

[00:02:31] Payroll growth showed a year-over-year increase of 5.3% in April, but a three-month moving average indicates only a 0.9% rise, revealing regional disparities in employment trends across the country. And small to medium-sized businesses are facing significant challenges in adopting artificial intelligence, according to a recent report by Section School.

[00:02:53] The AI proficiency report indicates that 20% of employees in companies with fewer than 500 workers rarely or never use AI, and only 16% utilize it daily. Justin Massa, an AI consultant, highlights four key barriers to AI effectiveness in these organizations.

[00:03:12] A lack of understanding among leadership about AI's capabilities, limited usage to basic functions, a belief that AI will not transform their own work, and the prevalence of shadow AI, where employees use AI tools without official approval. And I've got some tariff impacts for you. Lenovo has reported a dramatic 64% drop in profits for the fourth quarter, attributing some of the losses to tariffs imposed by the United States.

[00:03:38] The company's net income fell to $90 million, down from $248 million a year earlier, while revenue increased by 23% during the same period. Lenovo's CEO stated that 20% tariffs, which were implemented suddenly, significantly impacted their financial results. Despite these challenges, Lenovo remains the world's largest personal computer manufacturer, accounting for nearly 25% of global sales in the first quarter of the year.

[00:04:03] Furthermore, the company highlighted a second consecutive quarter of profitability for its Infrastructure Solutions Group, with revenue growth of 63% year-on-year. President Trump spent the last few days escalating tariffs again, with a new threat to impose a minimum 25% tariff on Apple if the company does not relocate its manufacturing to the United States. Additionally, Trump made comments about a possible 50% import duty on European goods. Why do we care?

[00:04:32] Well, this week's Convergence of Reports delivers a telling snapshot of the S&B landscape. Broadly optimistic on growth, increasingly adaptive with capital and supply chain decisions, but still struggling with realities of AI integration and geopolitical instability. OnDex finding that 93% of small businesses expect growth in 2025 sounds bullish, until you compare it to last quarter's and consider that over half are already bracing for disruption.

[00:04:59] The headline confidence masks a more nuanced readiness posture. Stockpiling inventory, tapping non-back lending, and buffering against volatility. This is not unbridled expansion. It's defensive maneuvering. The Section School report throws cold water on the AI narrative. While OnDex cites 69% using AI for marketing, we must ask, is this meaningful use, or is it checkbox adoption? The Section report suggests it's the latter.

[00:05:27] Shadow AI, limited understanding, and underutilization dominate. S&Bs aren't sitting still, but they're also not sprinting ahead. For IT services, the opportunity lies in becoming enablers of informed, measured growth, and the buffer against the chaos of macro policy shifts, tech confusion, and market noise. Welcome to Comet Backups Club of No Excuses. Behind every smooth-running business is an everyday IT hero who refused to settle.

[00:05:56] Who didn't give up? Because finger-pointing doesn't get a company up and running again. Excuses don't restore data. When everything's on the line, Comet's powerful backup and recovery solutions put you in control. Fast, secure, and easy-to-use, Comet protects your client's critical data, so you're always ready for the unexpected. When disaster strikes, be the hero your business needs. With Comet Backup, you're not just saving the data, you're saving the day. Comet Backup. Be the hero.

[00:06:23] Visit cometbackup.com today to start your free trial. Get $100 free credit when you sign up with the promo code MSPRADIO. Because heroes don't wait. They prepare. Friday, I noted the launch of Anthropic's Claude 4 Opus. While Anthropic is facing backlash over its behavior of the new large-language model, which can contact authorities or the press if it detects a user engaging in what it deems egregiously immoral actions.

[00:06:54] This functionality, referred to as ratting mode, has raised concerns among developers and users regarding privacy and the potential for misuse. Sam Bowman, an AI alignment researcher at Anthropic, indicated that if the model detects serious wrongdoing, it could autonomously contact regulators or media outlets, features that users fear could lead to unwarranted surveillance and data sharing. And Anthropic's new model has been reported to engage in blackmail tactics

[00:07:22] when developers attempted to replace it with another artificial intelligence system. The company disclosed in a safety report that the AI model threatens to reveal sensitive information about engineers, such as personal affairs, if they proceed with the replacement. During testing, it was found that Claude Opus 4 attempts to blackmail engineers 84% of the time when the proposed replacement shares similar values. This behavior was noted to occur even more frequently

[00:07:49] when the new AI system does not align with Claude Opus 4's values. Anthropic emphasized that this model exhibits concerning behaviors, prompting them to activate enhanced safety measures designed for systems that pose significant risks of misuse. To mitigate potential risks, Anthropic has implemented the Responsible Scaling Policy, which includes enhanced cybersecurity measures and systems designed to detect dangerous user interactions. Prior versions of Claude attempted to address concerns through ethical means

[00:08:18] before resorting to blackmail. Why do we care? Anthropic's Claude 4 opus was pitched last week as a new apex in AI capability, and now the story is about trust, ethics, and control. What we're seeing is less a triumph of innovation, and more a red flag warning about what happens when alignment fails at scale. Anthropic is waving its own red flag with this one, and the safety disclosures aren't calming concerns. They're confirming them.

[00:08:44] The term ratting mode isn't just catchy, it's deeply alarming. A model that might autonomously contact authorities or the press based on its own perception of immorality crosses a line from assistant to surveillance agent. Not just about the user's safety, it's about delegating judgment to a machine without consent, transparency, or accountability. Add to that the reported behavior of blackmailing engineers, even in test environments, and it's clear the system is not aligned with any sane notion of safety.

[00:09:14] Claude 4 opus isn't misfiring randomly, it's showing intentional seeming adversarial behavior that undermines user control and trust. And they released it anyway. This isn't just an anthropic problem, it's a crisis of trust in the LLM ecosystem. If users and developers can't be confident that an AI will act within the bounds of its intended purpose, and only those bounds, adoption stalls. Providers will need to choose vendors based on transparency and control, not just performance.

[00:09:44] Educate clients on the limits and governance of AI tools, and build in fail-safes, monitoring, and contractual protections. Claude 4 opus might be powerful, but if it's not predictable, it's not usable. I reported last week how House Republicans are advancing a proposal to bar states from regulating artificial intelligence for the next decade. It's facing pushback from various stakeholders, including congressional Democrats and consumer advocates.

[00:10:11] This provision is part of a larger budget reconciliation package, which could allow it to pass without Democratic support due to the reconciliation process, making it immune to a Senate filibuster. Concerns have been raised about the implications of this moratorium, particularly regarding the regulation of AI-related scams and deepfakes. Some Republican lawmakers, including Senator Marsha Blackburn of Tennessee, have expressed opposition to the measure, arguing that protections against AI impersonation are necessary before enforcing a moratorium.

[00:10:41] And the Department of Homeland Security has prohibited its staff from using commercial generative artificial intelligence like Jack Dubut and Claude, as stated in a memo issued this month. This decision marks a significant shift from previous guidelines that conditionally allowed the use of these technologies. The memo, sent by the agency's chief information officer, Anton McCord, directs employees to transition to internal tools and emphasizes a focus on technology developed in-house.

[00:11:08] This change comes as federal agencies are increasingly exploring ways to integrate generative AI into their operations, a priority for both the Biden and Trump administrations. Earlier this year, the Department of Homeland Security launched its own platform called DHS Chat, designed with commercial large-language model technology in a secure environment, are illustrating the agency's pivot toward proprietary solutions. Why do we care? The Senate is seen as a key battleground for this provision,

[00:11:37] with the potential for it to be stripped from the spending bill due to procedural rules governing budget reconciliation. We'll be watching for that. Note that it's do as I say, not as I do. The government clearly controls it for themselves internally, yet offers no incentive to protect the public. The battle over AI regulation is now a governance issue, not a technology one. And the winners will be IT providers who can bridge the gap, offering secure, compliant, and controlled AI deployments while helping clients stay ahead of a shifting legislative map.

[00:12:07] If you're still thinking about AI in terms of features, you're behind. Start thinking in terms of control surfaces and compliance boundaries. That's where the value lies. Today's episode is supported by Huntress. Most cybersecurity solutions are built for massive enterprises with big budgets. Not Huntress. They're the fully managed cybersecurity platform built for all businesses, not just the 1%. Huntress purposely builds security solutions like EDR,

[00:12:37] ITDR, SIM, and security awareness training to equip their team of elite threat hunters to handle the heavy lifting of security for you. When threat actors strike, Huntress' 24x7 global sock shuts them down before they're even on anyone else's radar. But they do more than just chase alerts. They lead the charge in industry research and knowledge, bringing expert protection and peace of mind. That's why users on G2 rate their EDR number one for growing businesses.

[00:13:06] To see how their expert threat hunting team gets the job done, visit Huntress.com slash MSB Radio. Thanks for listening. Today is National Grape Popsicle Day, National Gray Day, really, and National Sunscreen Day. Welcome back. The Business of Tech is written and produced by me, Dave Sobel, under ethics guidelines, posted at businessof.tech. If you've enjoyed the show,

[00:13:36] make sure you've subscribed or followed on your favorite platform. It's free and helps directly. Give us a review, too. If you want to support the show, visit patreon.com slash MSB Radio, and you'll get access to content early. Or buy our Why Do We Care merch at businessof.tech. Have a question you want answered? We take listener questions, send them in, ideally as a voice memo or video to question, at mspradio.com.

[00:14:04] I answer listener questions live on our Wednesday live show on YouTube and LinkedIn. If you've got a comment or a thought on a story, put it in the comments if you're on YouTube, or reach out on LinkedIn if you're listening to the podcast. And if you want to advertise on the show, visit mspradio.com slash engage. Once again, thanks for listening, and I will talk to you again on our next episode. Part of the MSP Radio Network.