News
The recent release of Anthropic Mythos is a wake-up call for the tech industry – and the fact that Anthropic themselves chose not to release it publicly speaks volumes about the level of risk we have now reached. AI agents have evolved from chatbots with upgraded capabilities to effective employees with database access, API keys, and system privileges.
However, the security protecting them is built on the same strategy that failed to stop ChatGPT jailbreaks in 2023. And this time, there’s no human to review an agent’s output, just an autonomous agent carrying out commands in a silo.
AI agents are reshaping enterprise systems and the way work gets done. Securing them requires an equally fundamental shift in thinking. Ultimately, now that agents act independently, resilience must be rooted in foundational controls, including hardware-level and lower-stack security, to be ready when the higher-level safeguards fail.
How AI agents expand the attack surfaceBefore agentic AI, the biggest AI risks were bad recommendations, inappropriate responses, and conversational data exposure. Human oversight acted as a safeguard for every action, and AI systems operated without direct access to sensitive information. The primary concern was reputational damage rather than risks to underlying infrastructure.
When Anthropic released the Model Context Protocol (MCP) in November 2024, it established a standardized framework that allows AI agents to connect to databases, file systems, and enterprise tools. But within eight months, a critical vulnerability emerged (CVE-2025-49596, CVSS9.4), triggering emergency security responses across the industry.
The risk came from four factors working together. Autonomy means agents can decide and act without human review. Privileged access gives them credentials, tokens and file system permissions. Machine-speed execution leaves little time for human intervention. And cross-system reach means one compromised agent can move across connected environments.
Together, these factors expanded the attack surface far beyond what traditional security controls – even AI-enabled ones – were built to manage.
Why software-only defences keep falling shortThe industry is moving quickly to secure AI agents, but the response largely mirrors a familiar approach: adding more layers of software. Most companies are focusing on two main layers: input guardrails – implementing more software tools designed to stop malicious instructions from ever reaching AI agents, and permissions and monitoring – limiting what compromised agents can access.
It’s the same strategy the industry had relied on for decades: deploy quickly, remain agile, and address vulnerabilities as they emerge. Both methods operate inside the software trust boundary.
But history shows this approach often ends the same way: with the need for hardware-layer protections. In the 1990s and 2000s, network security responded to software exploits by deploying additional software layers. Breaches persisted until organizations eventually adopted hardware-enforced network segmentation.
The same pattern played out with endpoint security in the 2000s and 2010s. As malware evolved to bypass detection, the response was behavioral analysis, sandboxing, and endpoint detection and response. Yet more software. Breaches continued until TPM (Trust Platform Module) chips and hardware-enforced secure boot became widely adopted. Cloud security, in the 2010s and 2020s, followed a similar path.
A common lesson runs through each of these domains: when the software trust boundary is compromised, the hardware layer – where data actually lives – must be secured too.
The case for hardware-level securityThis time, we cannot afford to learn slowly. Agents are already being connected to the systems that business rely on for their daily operations. Incidents like the MCP critical vulnerability and recent reports of a data leak caused by a Meta AI agent show how quickly the risks can become real.
Guardrails, permissions, and monitoring are necessary, but they are insufficient, and they represent the security layers that history shows will eventually be bypassed. Effective defense requires a third layer – one that exists beyond the software trust boundary and provides oversight at the hardware level, where sensitive data is ultimately stored and processed.
Hardware Root of Trust serves as the final security barrier, helping contain breaches before they escalate into a full system compromise. As the number of companies using AI agents continues to grow, security needs to move deeper than the application layer.
The industry has already learned that software alone cannot secure complex systems – it should not wait for a major compromise to learn the same lesson again.
We've featured the best endpoint protection software.
This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.
The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit
Now that Assassin's Creed Black Flag Resynced has been out in the wild for several weeks now, many of us die-hard AC fans have devoured every corner of its world and hoovered up every collectible and secret to dive deep into the lore once again.
It's a fantastic remake of one of the best Assassin's Creed games, and also has plenty going on in its stories, lore, quests, characters, and world that can be fun to test your knowledge on.
So I've built this quiz to do exactly that! Featuring questions on Black Flag and the Resynced remake, below are 15 questions that cover characters, the setting, links to other games, and more!
I've tried to make this Assassin's Creed Black Flag and Resynced quiz a bit tricky in places, and going a little deeper into lore here and there, so I hope it's the right balance of a test of knowledge combined with a few intriguing melon scratchers — including one that can't be confirmed, but for the life of me I am convinced is correct.
Good luck!
How did you find the quiz on Black Flag? A bit niche in places or a good test? Let me know in the comments, and feel free to let me know what other quizzes on games we could try!
A small business owner receives a call from a customer.
An email that appeared to come from the business led the customer to a fraudulent website, and their personal information may have been compromised.
What makes situations like this so damaging is that the owner never knew the risk existed.
The domain used in the attack had been registered for a campaign years earlier and left quietly active, sitting outside anyone's management, until someone else found a use for it.
Situations like this rarely begin with a major security breach. More often they start with something small: a forgotten domain, an outdated email configuration, or a digital asset nobody realized was still active.
AI makes finding those unnoticed weaknesses all too easy for attackers. According to KnowBe4's 2025 Phishing Threat Trends Report, 82.6% of phishing emails now show some use of AI, a 53.5% increase year-over-year. Attackers are adopting the same automation businesses use to improve efficiency, and the speed advantage has shifted. The good news is that businesses can use the same advances in AI to identify many of these risks before attackers do.
But the speed of detection is only part of the challenge. The harder problem is visibility. You can't secure what you don't know you own.
More and more, small businesses manage domain names, websites, email systems, social channels, and third-party tools, and as those layers expand, the gaps between ownership and oversight become easier to miss.
Forgotten domains are a common example. Domains created for promotions, campaigns, or discontinued services often stay active long after their purpose disappears. Left unmanaged, they become blind spots that attackers exploit through phishing, impersonation, and brand abuse, and they can quietly erode credibility well before any breach, since a domain that no longer resolves correctly signals neglect to customers and machines alike.
Simply put, many business owners no longer have a complete view of the digital assets they own or the vulnerabilities that come with them.
Security needs to be embedded, not bolted onSmall business owners are focused on serving customers, growing revenue, and running their businesses. They are not thinking about DNS records, certificate renewals, or dormant subdomains during their day. Nor should they have to.
But many do not have the option. According to VikingCloud's 2026 research, 84% of SMB owners manage cybersecurity themselves, often without dedicated training or expertise.
The most effective security strategies are built into the infrastructure which businesses depend on every day, rather than added after problems arise. There are three layers where this matters most: the domain, which serves as a business' identity online; the website, where customers form opinions about credibility and trustworthiness; and email, which remains one of the most important channels for customer communication and one of the most common targets for impersonation and fraud.
Embedding security into the solutions businesses already use helps them maintain visibility and confidence without constant manual oversight of all of those moving parts.
Trust is now measured by both people and machinesToday’s cybersecurity conversation always comes back to trust. That’s what SMBs want to win by securing their online business. It is the engine of their success.
We know that trust is one of the most important competitive advantages a business can have: according to McKinsey reporting on digital trust, 40% of consumers have completely pulled their business from a company after discovering the organization was reckless with customer data, and 10% of consumers will cut ties with a brand immediately upon learning of a data breach, regardless of whether their own personal information was actually compromised or stolen.
For years, trust online was primarily a human judgment. Customers visited a website, received an email, or interacted with a brand, and decided whether it appeared credible. Today they still make those decisions, but they are no longer the only ones making them.
Search engines, AI assistants, and automated systems increasingly evaluate trust signals on behalf of users. Roughly two-thirds of Google searches now end without a click, according to Similarweb clickstream data analyzed by SparkToro. Trust is no longer just a customer's perception; it is becoming part of how businesses get discovered.
Credibility is no longer determined solely by what customers see. Domain resolution, certificate validity, email authentication records, and the consistency of businesses’ online presence all feed into the assessments that influence search rankings, AI-generated recommendations, and discovery across the platforms customers use every day. A business that doesn’t deliver on these fronts may be overlooked long before a customer ever decides whether to trust it.
Trust and security are now competitive infrastructureFor decades, businesses viewed security as a defensive function, meant to reduce risk and respond to threats. That perspective is changing.
Trust and security now influence customer acquisition, retention, reputation, and long-term growth. In the AI era, trust is no longer just a security outcome. It is a business strategy. As AI accelerates both innovation and risk, customers have become more selective about who they engage with and where they share their information. Credibility is difficult to earn and almost impossible to buy back once lost.
At Network Solutions, we have spent decades helping businesses establish and protect their digital identities. One lesson remains consistent: investing in trust early creates advantages competitors struggle to replicate.
Businesses that stand out in the years ahead won't simply adopt more AI. They'll build trust into every layer of their digital presence. The business owner who took that call from a customer deserved a better security infrastructure, not a better incident response after the fact.
Technology will continue to evolve, and so will the threats. Trust will only become more valuable. The businesses that thrive won't simply adopt more AI; they'll build stronger foundations for trust. That's where our industry needs to go next.
We've listed the best free website builders.
This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.
The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit
- Trump levies huge tariffs on 'sensitive' drones, such as those heavier than 25kg
- Tariffs have also been applied to a range of other countries, such as the EU, UK, Japan, and Taiwan
- Trump says the tariffs will help strengthen national security while also bolstering the domestic drone market
President Trump has levied a new set of tariffs against imported drones and their components for national security reasons.
For ‘particularly sensitive’ drones, such as those over 25kg, those with docking stations, or those with thermal imaging capabilities, the tariffs are 100%.
“President Trump’s drone tariff program will protect the national security of the United States and its defense and defense-adjacent industrial base, supporting and creating American jobs,” the White House fact sheet said.
Strengthening US industry through national securityDrones below the 25kg threshold will still be subject to a 25% tariff, with drones imported from the European Union, Japan, Liechtenstein, Republic of Korea, Switzerland, and Taiwan subject to a 15% tariff. The United Kingdom has a lesser tariff of only 10%.
The intended consequence of these tariffs is to strengthen the US domestic drone construction industry and the wider domestic drone market.
Outside of outright upping the cost of drones in the US as vendors pass on the costs of drones to consumers, the tariffs will also likely increase the costs of an already constrained supply of specialized drones in the US market.
Many drones imported from abroad, particularly those from China, are used for specialized purposes such as search and rescue and power line inspection. As the US does not have the dedicated industries and resources to quickly begin manufacturing these drones, there will likely be a huge increase of demand for a very limited supply.
The new tariffs also close a loophole allowing Chinese parts to be assembled into a drone in the US. The fact sheet states that drones for commercial and US military purposes “rely on foreign sources for critical Unmanned Aircraft System (UAS) components, which poses significant risks to U.S. national security and creates cybersecurity vulnerabilities.”
The tariffs are due to kick in on September 3 2026, giving the US drone market just 21 days to adjust or pay massively inflated import costs.
If you're a seasoned anime fan, you'll understand the unique sense of excitement that comes with binging a brand-new show — and it's exactly how I feel about Crunchyroll's breakout series, BLACK TORCH.
As of writing, we're now six episodes into the epic adventure, following high-schooler Jiro, who just also happens to be a ninja descendant who can talk to animals. He's rescued an injured black cat named Rago, who is actually a powerful mononoke, or demon spirit.
I'll let Crunchyroll explain the next part: "After Jiro is mortally wounded protecting Rago from hostile spirits, Rago fuses with him, creating a human-spirit hybrid. Jiro is then recruited by Japan's covert Bureau of Espionage into a special unit called 'Black Torch' to fight rogue mononoke."
It's a classic anime storytelling format with a new twist, and so far, its six episodes have successfully blended high-intensity action with a strong emotional core. Even the intro song feels a cut above the rest, and I'm not surprised that anime fans are lapping it up.
With six episodes left to go, what can we expect from the second half of BLACK TORCH? According to English dub cast AJ Beckles (Jiro) and Ryan Colt Levy (Ryosuke Shiba), you're best off expecting the unexpected.
'It's more of what we've had... but bigger'"Expect the unexpected," Beckles tells me. "It's more of what you've got, but bigger. As we reach the end of the season and these episodes start to come out, there more high-octane, serious moments. There's a lot to look forward to, for sure."
It's a vague response, but this doesn't surprise me. Even the above trailer showed almost no signs of what was to come in the actual series, which was a refreshing decision. Of course, we can expect the drama to follow the original manga, but this can't be assured as a given.
But has there been any pressure for either actor knowing that these episodes are airing to a pre-existing fanbase?
"I don't know if it's necessarily like a pressure... I like to think of it as a responsibility," Levy says. "No matter what, there are people out there that you are making this for, and where we may never get to have that face-to-face feedback, we have access. People will put so much passion into that.
"We approach this no matter what with the intention to make sure that people who love it feel seen, and feel like they are spending time with the characters that they want."
"We've been cast and It's a ginormous responsibility, because it could have gone to anybody," Beckles agrees. "So the fact that it went to us means that we have a duty to do a good job. We're not necessarily thinking about the end result. The focus should be on, 'Am I thinking as Jiro is thinking? Am I serving this character?' Because if I do so to the best of my ability, someone will resonate with that."
I've got no idea how I'm going to feel about the upcoming second half of BLACK TORCH, but hearing the sheer passion that the voice cast has for its craft fills me with hope — and the joy that my initial burst of anime excitement seems set to continue.