mg.co.za Media Bias



Source identity and scope: The sample is unmistakably the South African Mail & Guardian — one record names it directly: Regulator spokesperson “told the Mail & Guardian”   .

Coverage is almost entirely domestic South African, spanning courts, GBV policy, mining, sport, arts, and science, dated late September–October 2026.

Topic selection: The sample splits into three observable clusters.

First, hard news and court reporting that is largely disciplined and attribution-based: the Malema leave-to-appeal brief is flat and procedural with named judges and a quoted order   ; the Sibiya bail report attributes all allegations to the state and notes his denial   ; the Ramaphosa GBV announcement is restrained despite lurid subject matter     .

Second, a large cluster of admiring profiles and advertorial-style content — the single most visible bias pattern.

Profiles of Khunou   , Fitchett   , Dzerefos   , Mafoko   , Sithebe   , and Motau   are uniformly celebratory with unverified claims (e.g., Motau's credentials have “no source cited”   ).

Commercial content reads as promotion: the Raslouw Hospital piece is self-sourced marketing language   , the Takealot/SETA launch uses “ambitious” framing without scrutiny   , the Green Point property piece is bullish developer copy   , the jazz festival recap is a “rhapsodic” rave   , and the DXY explainer embeds a broker link   .

Third, opinion/advocacy: a signed Gumede column attacking ANC cadre deployment   , a strongly worded commentary calling US UN arrears an “assault”   , and a pessimistic systemic analysis of ward councillors   .

Worldview: Evidenced leanings are progressive/community-rights, pro-science, pro-multilateralist, and institutionally reformist. The Tendele coal story foregrounds community loss while still publishing the company's detailed rebuttal   ; the pesticide piece sides with scientists urging EU precaution   ; GBV coverage prescribes expanded state-funded survivor services   ; UN peacekeeping is framed as a victim of US funding cuts   . No counterexample conservative framing appears in the sample.

Factual reliability: Mixed but generally decent on hard news — visible sourcing, hedges (“reported/alleged”)   , and uncertainty markers   support credibility.

Weaknesses: loaded openings that outrun attribution (“The extortionists” before evidence   ; “escape route hits a dead end”   ) and headline/body mismatches   .

AI authorship: Low probability for core news; human bylines, first-person reporting (“takes my call while driving”   ) argue against it. Some boilerplate-heavy promotional profiles     are weakly consistent with templated writing, but this is a probabilistic hypothesis only.

Propaganda techniques: No state-style propaganda evidence.

Observable techniques are softer: celebrification in profiles     , loaded language     , and advertorial blurring     .

Top 3 values: 1) Accountability/institutional integrity       ; 2) Community and lived-experience voices       ; 3) Scientific/empirical authority     .
Bottom 3 values: 1) Commercial independence (frequent advertorial content)       ; 2) Verification discipline in soft content       ; 3) Emotional restraint in feature writing     .



Helium Bias: Analysis is limited to 40 recent records (late Sept–Oct 2026), possibly selected toward high-bias items; I cannot observe unpublished stories.

Topic counts are my own categorizations of record summaries, not verified archives.

Source identity rests on one explicit mention and promotional-content inferences are probabilistic; hidden motives are not directly evidenced.

Automated source summary · Updated October 04, 2026 · Not human reviewed. Check recent article panels for claim-level evidence when available.




Use the Data in AI All Sources

mg.co.za Bias Profile

Weighted source-level patterns from recent analyzed coverage. Open recent articles below to inspect score-specific evidence and limitations when available.

🔵 Liberal <—> Conservative 🔴-14

🗽 Libertarian <—> Authoritarian 🚔20

📉 Bearish <—> Bullish 📈17

😩 Pessimistic <—> Optimistic 🌞14

💡 Boring <—> Interesting20

😨 Fearful36

📞 Begging the Question20

💭 Opinion100

🗳 Political36

Oversimplification24

🏛️ Appeal to Authority32

👀 Covering Responses40

😢 Victimization16

😤 Overconfidence50

🗑️ Spam30

❌ Low Credibility <—> High Credibility ✅26

🧠 Rational <—> Irrational 🤪-24

🤑 Advertising57

💔 Low Integrity <—> High Integrity ❤️26

🪨 Low Intelligence <—> High Intelligence 🦉52

🔍 Truth-seeking <—> Delusion 🌀-50

🔬 Scientific <—> Superstitious 🔮-32

👤 Individualist <—> Collectivist 👥20

🎲 Speculation32

🐍 Manipulative87

Subtle dimensions

🧢 Populist <—> Elitist 🎩-4

🗞️ Objective <—> Subjective 👁️ 2

🚨 Sensational0

📝 Prescriptive0

🏴 Anti-establishment <—> Pro-establishment 📺2

🦊 Anti-Corporate <—> Pro-Corporate 👔-4

How to interpret source scores →

Average social shares per article 0





mg.co.za Recent Articles



Sort By: