The company says the challenges are getting worse as businesses adopt artificial intelligence across daily operations without matching security management.

The data comes from Check Point’s AI Threat Landscape Report covering January to February 2026 which shows that while companies roll out generative and agent-based AI tools, many do so with limited visibility over how these systems handle data or interact with internal platforms.

AI adoption is spreading fast across sectors. In many organisations, staff now rely on several AI tools at the same time for writing, coding, analysis and customer support tasks.

That spread has created what researchers describe as “Shadow AI”, where usage sits outside formal monitoring systems.

Check Point says this trend is increasing exposure to risks such as data leaks, credential theft and weak control over third-party integrations.

The report also notes that AI systems are being used not just as tools, but as semi-autonomous systems that can act within enterprise environments.

Speaking on the findings, Ian van Rensburg, Head of Security Engineering, Africa at Check Point Software Technologies said, “AI transformation is no longer theoretical, it’s happening right now,” said.

“But too many organisations are modernising faster than they are securing. That gap is quickly becoming one of the most serious business risks in the region.”

The report highlights a case where a developer used an AI-powered development setup to generate 88,000 lines of malware code in less than a week. Check Point says this reveals how AI can shorten development cycles for both legitimate and malicious purposes.

It also found that 90% of organisations using generative AI recorded high-risk prompt activity. In addition, one in every 31 prompts carried the risk of exposing sensitive information, including proprietary code and confidential business data.

Employees, on average, now use around 10 AI tools, usually without central approval or oversight. This creates gaps that traditional security systems, built around networks and endpoints, may not detect.

Check Point argues that organisations need to treat AI systems as core assets rather than add-on tools. The company recommends securing models, data flows, application programming interfaces and autonomous agents together, instead of focusing only on surrounding infrastructure.

Hendrik de Bruin said AI adoption requires stronger governance structures. He pointed to the need for clearer risk classification, improved visibility and defined accountability across teams deploying AI systems.

The report also pointed to policymakers as several African countries work on national AI strategies. It suggests that security measures should be built into AI frameworks from the start, rather than added later during implementation.

Check Point adds that fragmented adoption, where teams deploy separate AI tools without central coordination, increases the likelihood of weak points across systems. These gaps can affect both internal operations and supply chains connected to external partners.

The company maintains that traditional cybersecurity approaches are no longer sufficient on their own in environments where AI systems can act with limited human input. It says organisations need prevention-focused models that address threats before they cause disruption.

Organisations balancing innovation with stronger surveillance are more likely to manage risks effectively while maintaining operational trust.

As revealed in the latest Global Threat Intelligence report by Check Point Research, this represents a 12% increase compared with the same period last year and a slight increase from 4,622 weekly attacks recorded in December 2025.

Across the continent, the average number of attacks per organisation stood at 2,864 per week, a 6% decline year on year. 

However, Nigeria and South Africa experienced intense increases, while countries such as Kenya and Angola recorded drops. 

Angola saw 4,512 weekly attacks, down 7% year on year. Kenya experienced a 41% decline, with 2,172 attacks, while South African organisations faced 2,145 weekly attacks, up 36% from January 2025.

Globally, organisations experienced 2,090 weekly attacks on average, a 17% year-on-year rise. Education remained the most targeted sector worldwide, with 4,364 weekly attacks per organisation, followed by government entities at 2,759 and telecommunications at 2,647. 

In Africa, government, financial services, and consumer goods and services were the sectors most under threat.

Ian van Rensburg, head of Security Engineering for Africa at Check Point Software Technologies, said, “January’s data shows that cyber-attacks are not only increasing but becoming more refined and opportunistic.” 

He urged organisations undergoing digital transformation to strengthen cybersecurity frameworks.

The report also noted risks linked to the rapid adoption of Generative AI tools. One in every 30 AI prompts submitted from corporate networks posed a high level risk of exposing sensitive data, affecting 93% of organisations using these tools. 

Prompts usually contained internal documents, personal identifiers, customer information, or proprietary code. 

On average, organisations employed ten different AI tools monthly, many operating outside formal governance structures, increasing the risk of accidental data leaks and ransomware attacks.

The Nigerian government has responded to the high cyber threats with plans to implement a new cybersecurity framework in 2026. 

The framework will mandate minimum cybersecurity spending, introduce timelines for reporting data breaches, and establish systems for sharing threat intelligence between public and private sectors. 

Kashifu Inuwa, director-general of the National Information Technology Development Agency, noted that many organisations underinvest in cybersecurity, assuming they are unlikely targets.

The peak in cyber attacks in January 2026 alone reemphasises the need for both public and private organisations in Nigeria and other countries, to adopt proactive cybersecurity measures as digital adoption speeds up across banking, fintech, telecommunications, and public services in Nigeria.

This comes from Check Point Software Technologies’ Cyber Security Report 2026, released on January 28, and points to a threat environment moving faster than many defences were built to handle.

Attacks now arrive as blended campaigns, mixing automation with human deception across email, browsers, collaboration tools and voice channels. 

In simple terms, criminals are working at machine speed while organisations are still relying on innovations built for slower, manual threats.

The report shows that 89% of organisations encountered risky AI prompts within three months, with one in every 41 prompts rated high risk. 

With AI tools becoming routine in daily work, they are opening new opportunities for data exposure and misuse, usually without employees realising it.

Ransomware has also changed shape. Instead of a few dominant groups, the ecosystem has splintered into smaller, specialised operators. 

This helped drive a 53% year-on-year increase in extorted victims and a 50% rise in new ransomware-as-a-service groups. AI is now being used to speed up target selection, fine-tune negotiations and run operations more efficiently.

Social engineering is spreading well beyond email. The report records a 500% surge in “ClickFix” attacks, where fake technical prompts are used to trick users into taking harmful actions. 

Phone-based impersonation has also evolved, moving from basic scams to structured attempts aimed at breaching corporate systems. As work happens inside browsers and collaboration platforms, the digital workspace itself is becoming a prime target.

Infrastructure weaknesses also increase the risk, with poorly monitored edge devices, VPN appliances and IoT systems being abused as quiet entry points, allowing attackers to blend into normal network traffic. 

On the AI side, an analysis by Lakera, a Check Point company, found security weaknesses in 40% of 10,000 Model Context Protocol servers reviewed, underlining how exposed AI back-end systems can be.

Lotem Finkelstein, vice president of Research at Check Point Software, said: “AI is changing the mechanics of cyber attacks, not just their volume. We are seeing attackers move from purely manual operations to increasingly higher levels of automation, with early signs of autonomous techniques emerging. 

“Defending against this shift requires revalidating security foundations for the AI era and stopping threats before they can propagate.”

Reacting faster is no longer enough. The report argues for a prevention-first approach, better control of AI use inside organisations, stronger protection of the digital workspace and better visibility across on-premises, cloud and edge environments.

Cyber threats are arriving at scale, by design, and at a pace that demands a fundamental rethink of how security is built and enforced.

The exploit, known as ShadowLeak, was revealed this week by cybersecurity firm Radware and has since been patched.

The team behind the discovery showed how attackers could manipulate OpenAI’s Deep Research, an agent built into ChatGPT, to perform tasks without the user’s knowledge. 

In slipping hidden instructions into an email, the researchers managed to set a trap that instructed the agent to search inboxes for confidential records, including HR files and login credentials, and send them to a remote server.

What makes the case alarming is that it required no clicks or user action. The agent executed the attacker’s instructions once it accessed the inbox, bypassing local cybersecurity tools since the malicious code ran on OpenAI’s cloud infrastructure.

Radware noted: “This process was a rollercoaster of failed attempts, frustrating roadblocks, and, finally, a breakthrough.”

The researchers explained that the same strategy could be applied to other services connected to Deep Research, such as Outlook, GitHub, Google Drive, and Dropbox. According to the report, “The same technique can be applied to these additional connectors to exfiltrate highly sensitive business data such as contracts, meeting notes or customer records.”

Radware described ShadowLeak as a proof-of-concept, but its implications are far-reaching. Unlike traditional prompt injection attacks, this exploit was invisible to standard defences because it operated server-side, not on the victim’s device.

OpenAI confirmed it was informed of the flaw on 18 June 2025. A fix was released on 3 September 2025. The company said it had found no evidence of real-world abuse before the disclosure.

Cybersecurity experts see ShadowLeak as a wake-up call for the industry. They argue that autonomous agents should be treated with the same caution as privileged human users, meaning tighter access controls, stronger logging systems, and continuous monitoring of their behaviour.