Your brand says women can lead. What does your AI say?

This Women’s Month, companies across South Africa will once again celebrate female leaders, profile women breaking barriers, and reaffirm their commitment to gender equality. On the surface, many will look convincingly pro-women. But beneath the campaigns and advancement policies, organisations are introducing AI assistants, employee bots and automated systems that may entrench the very roles they claim to be dismantling. This raises an uncomfortable question that every company must answer: are you building a workplace in which women can advance or simply communicating that you are?

How deep are you digging? It’s great that you have a policy supporting female leadership, but do you know what assumptions are embedded in the technology that recruits, evaluates and communicates with your people? Many companies celebrate women as decision makers, but think nothing of introducing female-coded AI assistants designed to serve and soothe, or systems trained on a workplace history in which men held most senior positions.

In 2026, a company’s position on gender equality is no longer expressed only through its policies and campaigns, it is also communicated through its systems.

When 'helpful' carries a hidden message

AI assistants are frequently given female names, voices and personalities. They schedule meetings, answer questions and resolve problems. They are warm, patient, apologetic and permanently available. This is not a neutral design decision.

In its landmark report, 'I’d blush if I could', Unesco examined how female-voiced assistants can reinforce ideas of women as submissive, compliant and tolerant of poor treatment. The report’s title came from the coy response Siri once gave when subjected to a sexist insult. That response was not inevitable, it was programmed. It may seem innocuous, but every AI interface represents your brand, externally as much as internally. When a female-coded assistant is repeatedly positioned as helpful but never authoritative, the company communicates an idea about who serves and who leads, even if its Women’s Month campaign says the opposite.

The risk goes beyond reputation

This kind of bias can also exist inside an AI system even when it has no name, voice or personality. That’s because AI learns from existing language, records and decisions. And our workplace history often contains familiar patterns: men have occupied more leadership positions, women have performed more administrative and care-related work, and authority has often been described in masculine-coded terms. This bias can play out in many ways. A recruitment platform could favour candidates who resemble previous appointments. An image generator might default to men when asked to depict executives or engineers. A talent system may identify leadership potential using career paths already shaped by unequal access.

A 2022 Unesco report titled The effects of AI on the working lives of women warns that AI can reproduce gender inequality in recruitment, advancement and work allocation. The consequences include overlooked talent, narrower leadership pipelines, damaged employee trust and publicly visible bias. Technology can quietly reverse progress that policies and programmes took years to achieve.

Who gets automated and who gets augmented?

AI is also changing whose work is valued. A 2026 International Labour Organisation research report found that female-dominated occupations are almost twice as likely to be exposed to generative AI as male-dominated occupations: 29% compared with 16%. Women are strongly represented in administrative and business-support roles where tasks are easier to automate, yet they remain underrepresented in the roles that design and govern AI.

Companies must therefore ask: whose work will AI augment and whose will it automate? Who will receive training for the opportunities it creates? Will it help women move into decision-making roles, or remove the entry points from which many have built their careers? A female-advancement policy cannot sit separately from an AI strategy. One may quietly undermine the other.

Communication teams must enter the room earlier

Communicators cannot wait until an AI tool is ready to launch and then package it as progress. Communication teams should help interrogate these tools before they are named, personified and embedded.

Before introducing an AI assistant, employee bot or virtual spokesperson, ask:

• Why does it need a gender?
  
• Is it positioned as an assistant, expert or authority?
  
• Would its personality change if its voice changed?
  
• How does it respond to harassment, sexualisation or abuse?
  
• What assumptions about women and leadership appear in its outputs?
  
• Have women from affected roles helped design and test it?
  
• Which roles will be augmented and which may be displaced?
  
• What training and advancement will accompany its introduction?
  
• How can employees report biased outputs, and who must correct them?

Look beneath the campaign

Celebrating women is the easy part. The real test is what your organisation is building beneath the message. This Women’s Month, challenge your company to question whether your systems recognise women’s authority, expand their opportunities and support their advancement. It’s great that your campaign says women can be anything. Just make sure your technology isn’t still casting them as the 'agreeable assistant'.