Most security awareness programmes still teach staff to watch for bad grammar, mismatched logos and generic greetings. In 2026, those signals are no longer reliable. Attackers now produce polished English and pixel-perfect fake invoices that lack the obvious warning signs employees were once taught to spot.
For SMB and mid-market organisations running lean IT teams, the gap between what training assumes and what attackers do has widened fast. Three shifts explain most of it: AI-written emails with detailed personalisation, QR codes replacing links, and attacks that skip email altogether in favour of Microsoft Teams, Slack and calendar invites.
AI is doing the personalisation work attackers used to do by hand
Attackers used to choose between scale and quality: blast generic templates to thousands of inboxes, or spend hours researching a single target. Large language models remove that trade-off. Given a target's LinkedIn profile, a recent company announcement or a scraped internal email thread, an LLM can generate a convincing message in seconds, in fluent, contextually accurate language.
Microsoft's threat intelligence team has been tracking this shift directly. In April 2026 the company documented an AI-enabled device code phishing campaign, one of several techniques its researchers flagged as evidence that credential theft methods keep evolving faster than static defences can follow[1].
The UK government's Cyber Security Breaches Survey found phishing remains the most prevalent type of breach or attack, hitting 38% of UK businesses and 25% of charities, and the most disruptive according to 69% of those affected. The survey reflects what we hear from IT leaders: phishing has become easier to execute, exactly what a lower barrier to entry from AI tools would produce[2].
In conversations with IT leaders, this comes up constantly: teams that spent years training staff to spot clumsy translation and spelling errors say those signals have largely disappeared, and even experienced IT professionals now struggle to tell a sophisticated attempt from a legitimate message.
QR codes are moving the attack off the corporate network
Microsoft recorded QR code phishing volumes jumping from 7.6 million in January 2026 to 18.7 million in March, a 146% increase over the quarter and the highest monthly volume the company has tracked in at least a year[3].
Part of this is technical: a malicious link gets scanned by corporate email security before it reaches an inbox, but most scanning engines still struggle to parse a link hidden inside an image. Part of it is behavioural: employees have grown used to scanning codes to authenticate a login or approve a payment, so a QR code feels routine rather than suspicious. Once scanned, typically on a personal phone, the session moves off the corporate network, routing around endpoint security, network monitoring and DNS filtering.
QR-based scenarios (an email claiming a user needs to scan a code to verify their account, for example) are now common enough to appear as their own category in training content, showing how quickly this moved from novelty to routine attack vector.
Attacks are following employees into Teams, Slack and their calendars
Perhaps the most significant shift for SMBs is that phishing is no longer confined to the inbox. Palo Alto Networks' Unit 42 found that phishing alerts originating from collaboration platforms made up 42% of all phishing alerts across its telemetry in the first four months of 2026, up from 30% in the preceding four months[4]. Attackers now regularly open contact by posing as IT support or a known vendor directly inside Microsoft Teams, where employees are more likely to assume messages are legitimate.
Two documented cases from Unit 42 show how this plays out. The group tracked as Cloaked Ursa (also known as APT29 or Midnight Blizzard) has used compromised accounts to send Teams messages containing credential-harvesting links disguised as legitimate Microsoft sign-in pages. Separately, a group tracked by Mandiant as UNC6692 impersonated IT helpdesk staff through Teams chat invitations sent from accounts outside the target organisation, convincing employees to approve fraudulent multi-factor authentication prompts[5].
In many organisations, Teams is configured by default to accept messages from any external tenant, which attackers exploit directly. Slack carries the same structural risk: any internal chat tool that employees trust by default is worth attacking.
This is exactly the gap modern security awareness platforms need to address. Training that only simulates email no longer reflects where employees are being targeted.
Calendar invites are following the same pattern. Check Point researchers uncovered a campaign that targeted over 300 organisations with more than 4,000 phishing emails in four weeks, all disguised as legitimate Google Calendar invitations from known contacts[6]. Because a calendar invite arrives as a routine scheduling notification rather than an email asking for action, it slips past the instinct most training has spent years building around suspicious links and attachments.
For SMBs, this is the real exposure: a programme that only simulates email leaves collaboration tools and calendar-based attacks uncovered, even as they become a growing share of what reaches employees.
How security awareness training needs to adapt
Closing that gap means training that follows attackers to where they now operate: simulations across email and collaboration platforms, scenarios that reflect QR codes and calendar invites rather than only suspicious links, and a programme that updates automatically as tactics shift, rather than an annual refresh of last year's threats.
Practice from Pistachio delivers continuous, fully automated security awareness training built around how employees are targeted in 2026.
Simulations across email and Microsoft Teams. Practice runs simulations directly in both channels, reflecting the same collaboration-platform risk Unit 42 and Microsoft are now tracking, not just inbox scenarios.
Role-based personalisation. Finance teams see invoice fraud and payment-request scenarios, IT teams see credential and vulnerability-themed attacks, matching how real attackers tailor messages by role and department.
Adaptive difficulty and frequency. If someone repeatedly falls for QR-code or Teams-based simulations, Practice automatically increases their exposure to those scenarios, while easing off training for employees who consistently show good judgement.
Reporting built for compliance. Evidence is collected automatically for NIS2, ISO 27001 and Cyber Essentials audits, cutting the manual reporting work lean IT teams would otherwise carry.
Phishing in 2026 does not fail because employees are careless. It fails because it has outgrown the channel and format most training still assumes. Attackers have already moved beyond the inbox. Security awareness programmes need to do the same.
See how Practice helps SMBs train employees across email, Microsoft Teams and other collaboration channels, without adding administration for already stretched IT teams. Email contact@pistachioapp.com to book your 15-minute demo.
