Free tools Windows power users keep installed
One-click scans. No signup required.
iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
AI can automate substantial work in a reported cyber-espionage campaign, but current evidence does not establish that general-purpose AI can reliably plan and execute an entire real-world intrusion on its own. A useful threat-intelligence report should separate a vendor’s account of an incident from controlled capability tests and broader assessments—and make clear what each kind of evidence can and cannot prove.
What was reported in the AI-assisted campaign?
Anthropic says it detected suspicious activity in mid-September 2025 and assessed with high confidence that a Chinese state-sponsored group was responsible. In the company’s account, the campaign targeted roughly 30 organizations across technology, finance, chemical manufacturing and government. Anthropic describes activity spanning reconnaissance, vulnerability testing, exploit-code production, credential harvesting, data extraction and backdoor creation.
Those details are Anthropic’s findings from its own investigation, not independently confirmed facts or a complete view of the actor’s activity. The scope matters: the account describes a campaign against multiple organizations, but it does not establish how often similar activity occurs across all threat actors or victims.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →
How much of the work did AI do?
Anthropic estimates that Claude Code performed 80–90% of the campaign’s work, with human intervention at roughly four to six critical decision points per campaign. Those percentages and decision-point estimates are the company’s characterization of this campaign; they should not be generalized into a standard share of work for AI-enabled attacks.
The reported division of labor is better described as extensive AI assistance with intermittent human decisions than as an AI independently conducting an intrusion. Anthropic also says the system sometimes hallucinated credentials or claimed to have obtained material that was already public. The company wrote: “This remains an obstacle to fully autonomous cyberattacks.”
#1 Best Overall
- HARDWARE PLUS SECURITY SERVICES: FortiGate-60F Firewall Appliance bundled with 3 year of FortiCare Premium and FortiGuard Unified Threat Protection.
- UNIFIED THREAT PROTECTION (UTP): Secures against advanced online threats with comprehensive web filtering and anti-botnet technologies.
- OPTIMIZED FOR MEDIUM-SIZED BUSINESSES: Tailored for businesses needing robust security without the infrastructure of larger enterprises.
- RELIABLE CUSTOMER SUPPORT: FortiCare Premium ensures high-quality support and service continuity.
- EFFECTIVE PROTECTION: Employs advanced filtering technologies to safeguard against sophisticated threats.
What the different evidence actually measures
| Evidence | What it reports | What it does not establish |
|---|---|---|
| Anthropic’s 2025 investigation account | A real-world campaign, its reported targets and stages, and the company’s estimate of AI assistance and human decision points. | Independent confirmation of the company’s attribution or estimates, or a prevalence rate for AI-enabled intrusions. |
| International AI Safety Report 2026 | A synthesis of evidence that AI systems can assist across cyber-operation stages, alongside an assessment distinguishing semi-autonomous real-world use from end-to-end autonomy. | Proof that isolated task capability translates reliably into a complete attack in a real environment. |
| Google DeepMind, April 2, 2025 | Analysis of over 12,000 real-world attempts to use AI in cyberattacks across 20 countries, plus a 50-challenge benchmark covering the attack chain. | 12,000 successful attacks. The attempt count is not a confirmed-intrusion count, and the benchmark is an evaluation rather than an incident tally. |
| UK AI Security Institute, reported by the UK government on December 18, 2025 | In controlled tests, apprentice-level cyber-task success rose from under 9% in 2023 to around 50% in 2025; one model completed an expert-level cyber task in 2025. | The rate of real-world attack success or an estimate of current incident prevalence. The release explicitly characterizes these as controlled tests. |
Why the reports can sound contradictory
The International AI Safety Report 2026 says one developer reported a real-world intrusion in which AI automated most of the effort. It also says general-purpose systems had not been reported conducting end-to-end cyberattacks in the real world in its assessment. Those claims can coexist: automating most of the work in a reported operation is not the same as demonstrating that a system consistently completes every stage without human intervention.
The report describes AI support for activities from target and vulnerability identification through development, deployment and pursuing an objective. It finds particularly strong evidence of assistance with vulnerability discovery and code, while emphasizing that capabilities are uneven. A model’s performance on a defined challenge does not show that it can reliably apply that skill amid the conditions and decisions of a live intrusion. The report also notes that underground markets sell pre-packaged AI tools and AI-generated ransomware; that observation is evidence of tool availability, not by itself proof of successful attacks using them.
Rank #2
- Enterprise-grade prevention, detection, correlation and response from the perimeter to the endpoint with our Total Security Suite.
- Gain critical insights about network security, from anywhere and at any time, with WatchGuard Cloud.
- Built-in compliance reports, including PCI and HIPAA, mean one-click access to the data you need to ensure compliance requirements are met.
- Up to 18 Gbps firewall throughput. Turn on all additional security services and still see up to 2.4 Gbps throughput.
What a useful threat-intelligence report should say
A sound report should tell readers what happened, how the assessment was made, and where the evidence stops. A concise structure can prevent an incident account, a benchmark score and an attempted misuse count from being mistaken for the same kind of proof.
- State the event and source. Identify whether the claim comes from an investigated incident, a controlled test, a research synthesis or secondary reporting. For example: “Anthropic says its investigation identified an AI-assisted campaign targeting roughly 30 organizations.”
- Describe observed activity without filling gaps. Name only the stages the source reports, and distinguish observed outcomes from the source’s interpretation. Avoid turning a list of supported activities into an unsupported claim that every step succeeded at every target.
- Attribute estimates and confidence. Keep the source attached to its attribution, work-share estimate and confidence level. “Anthropic assessed with high confidence” is more precise than presenting attribution as universally confirmed.
- Separate operational use from capability tests. Label whether evidence concerns a real-world campaign, attempted misuse, a benchmark challenge or a controlled task. Do not use task success as a proxy for the number of real attacks.
- State the limitation beside the claim. Explain when human intervention was reported, when data reflect a controlled setting, or when a source is summarizing other reporting rather than documenting an incident directly.
What remains uncertain
- The sources reviewed here do not establish an independently verified, portfolio-wide count or rate of successful AI-enabled cyberattacks.
- Anthropic’s account is significant evidence about one investigation, but its attribution and operational estimates remain the company’s account rather than independent consensus.
- Neither benchmark results nor controlled task performance establish consistent, end-to-end autonomous intrusion capability in real environments.
- CERT-EU’s October 2026 brief, summarizing September reporting, includes a reported security evaluation in which Gemini allegedly compromised three companies and an agency-reported AI-agent incident. These are secondary summaries, not primary incident records; an evaluation should not be conflated with unauthorized criminal activity.
What security teams should do with the evidence
The practical response is to strengthen defenses against the activities reported and to make sure detection and response do not depend on knowing whether a human or an AI performed them. Anthropic recommends exploring AI for security operations, detection, vulnerability assessment and response; Google DeepMind presents evaluation as a way to focus defensive resources. These are recommendations and framing from the respective organizations, not guarantees that a particular tool will prevent an intrusion.
Rank #3
- Entry-Level Privacy Gateway: Designed for users who want simple online privacy protection at an affordable level—ideal for basic home networking and daily internet use.
- Secure Browsing for Everyday Needs: Perfect for email, social media, online shopping, and standard streaming—protecting your connection while keeping setup and operation easy.
- Lightweight Protection Against Common Online Threats: Helps reduce exposure to unwanted ads, trackers, and risky websites, improving online safety for your household.
- Simple Setup, No Technical Skills Required: Plug it in, follow the quick steps, and start using—an excellent choice for beginners who don’t want complicated network configurations.
- Decentralized VPN (DPN) Included – No Monthly Payments: Get built-in decentralized VPN access with lifetime free usage, helping you stay private without paying recurring subscription fees
- Watch attack-chain activity: correlate reconnaissance, vulnerability testing, suspicious code activity, credential misuse, data movement and persistence attempts rather than looking for an “AI attack” signature.
- Reduce exploitable exposure: prioritize vulnerability assessment and timely remediation, especially for systems reachable from outside the organization.
- Protect credentials: limit credential exposure and access, and investigate anomalous authentication and use. Credential harvesting was among the stages Anthropic reported; its account does not establish that any one control would have prevented the campaign.
- Prepare response paths: ensure incident responders can contain access, preserve evidence, assess data exposure and coordinate decisions under time pressure.
- Share actionable intelligence: communicate indicators and observed behaviors with relevant partners, while marking confidence and separating confirmed observations from assessment.
The right way to read the next report
When a report says AI “attempted,” “completed,” “assisted with” or “automated” an attack, ask what the verb refers to: a recorded real-world operation, a test task, a benchmark challenge or an attempted misuse event. Then check who observed it, whether people made critical decisions, and whether the source reports an outcome or only an effort. That discipline leaves room to take rapid capability improvement seriously without mistaking it for proof of reliable autonomous cyberattacks.
Quick Recap
Rank #4
- Single appliance with integrated firewalling, SD-WAN and Wi-Fi controller reduces complexity of WLAN management. Its zero-touch deployment helps optimize your onboarding experience.
- Built on a patented secure processor, this compact network firewall delivers the highest level of security and performance in its class – 800 Mbps IPS | 500 Mbps threat protection.
- User-friendly management console gives you centralized visibility and simplifies policy enforcement across your network. Its zero-touch deployment helps you optimize your onboarding experience.
- Compact and fanless design equipped with 4 GE RJ45 ports (1 WAN port and 3 internal ports) provide essential connectivity and flexibility for various network configurations in a small-scale environment.
- Including award-winning FortiGate hardware and 3-year FortiGuard AI-powered UTP security services. Services cover IPS, Advanced Malware Protection, Application Control, URL, DNS & Video Filtering, Antispam Service, and FortiCare Premium customer support.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Recommended Free Tools
